Reconstructing sea surface temperature in the South - ePIC

Transcrição

Reconstructing sea surface temperature in the South - ePIC
 Reconstructing sea surface temperature
in the South Pacific using organic proxies
Dissertation zur Erlangung des akademischen Grades
eines Doktors der Naturwissenschaften
Dr. rer. nat.
an der Fakultät für Geowissenschaften der Universität Bremen
vorgelegt von
Sze Ling Ho
Bremerhaven, 2012
Gutachter der Dissertation
Prof. Dr. Ralf Tiedemann
Prof. Dr. Gesine Mollenhauer
Erklärung
Name: Sze Ling Ho
Anschrift: Sielstrasse 12, 27568 Bremerhaven.
Hiermit versichere ich, dass ich
1. die Arbeit ohne unerlaubte fremde Hilfe angefertigt habe,
2. keine anderen als die von mir angegebenen Quellen und Hilfsmittel benutzt habe
und
3. die den benutzten Werken wörtlich oder inhaltlich entnommenen Stellen als solche
kenntlich gemacht habe.
Bremerhaven, 20th September 2012
………………………….
(Unterschrift)
The most exciting phrase to hear in science, the one that heralds
new discoveries, is not ‘Eureka!’, but ‘That’s funny…’
Isaac Asimov (1920-1992)
Abstract
Sea surface temperature (SST), at the interface between the atmosphere and the ocean,
is an important element in the global climate system. Accurate estimates of past SSTs are
indispensable for studying global climate and for validating the numerical models used for
projections of future climate. SSTs prior to the instrumentation era could be reconstructed
using climatically sensitive biomarkers such as alkenones and glycerol dialkyl glycerol
tetraethers (GDGTs). Herein I use marine sediments, both core-tops and long piston cores, to
further examine the applicability of organic SST proxies derived from the aforementioned
biomarkers, with an emphasis on the relatively under-studied subantarctic Southeast Pacific.
The appraisal of these proxies is based on the correlation of the core-top proxy index values
with present-day climatological SST, and the comparison of the SST records inferred from
these proxies with other regional SST records.
The alkenone index (UK37 and UK’37) values in the South Pacific core-tops display
linear relationships with climatological World Ocean Atlas 2009 (WOA09) SST at low
temperatures (1 -12°C), with equally high r2 values (>0.93). These results suggest that both
alkenone indices are highly correlated to SST even at high latitudes, rendering them
appropriate for reconstructing SST in the subantarctic Pacific. However, these indices yield
different Pleistocene SST patterns for study sites in the subantarctic sector of the Southern
Ocean. Judging from the better structural fit of UK37 SST records with other subantarctic
surface proxy records, including foraminiferal 18O and SST records inferred from diatomand foraminiferal assemblages, it appears that the UK37 index results in more plausible paleo
SST records in this region, as opposed to the commonly used UK’37 index.
The GDGT index (TEX86 and TEX86L) values in the subpolar and polar core-tops
(with overlying SSTs of -2 to 17°C) are not highly correlated (r2 values < 0.3) to the annual
mean WOA09 SST. Plotting these indices against seasonal SSTs and water temperatures at
various hydrographic boundaries does not improve the correlation. Nevertheless, when these
data are combined with previously published core-top data (n = 630) spanning a SST range of
-2 to 30°C, the r2 values of these relationships improve to ~ 0.8, with a better correlation for
the TEX86 compared to TEX86L. The TEX86 calibration yields SST estimates that are in better
agreement with the WOA09 SST for subpolar and polar regions, rendering it a better SST
index in these regions. Meanwhile, applying the GDGT index on a Southeast Pacific sediment
core yields SSTs that are colder than those inferred from the alkenones, and the GDGTI derived interglacial SSTs are ~3°C colder than annual mean WOA09 SST. Given the lack of
seasonality in the archaeal abundance here, the cold-biased GDGT estimates probably reflect
subsurface temperature instead of SST.
The three SST records presented here allow for further understanding of the surface
oceanographic variability in the subtropical and subantarctic Southeast Pacific over the past
700 kyr. The extent of glacial cooling increases with latitude, up to ~8°C in the subantarctic
as opposed to the ~4°C in the subtropics. Intense cooling at high latitudes results in larger
latitudinal SST gradients during glacials than interglacials. The alkenone-inferred SSTs along
the latitudinal range of the Peru-Chile Current (PCC) imply massive equatorward migrations
of the Southern Ocean frontal systems and increased equatorward transport of subantarctic
waters owing to a stronger PCC during glacials. In addition, GDGT-derived temperatures
suggest enhanced subsurface warming during MIS 11 and 13 in the subtropical Southeast
Pacific, plausibly due to water column reorganization analogous to that occurring during
modern-day El-Niño conditions.
II Zusammenfassung
Die Meeresoberflächentemperatur, am unmittelbaren Kontakt zwischen Atmosphäre
und Ozean, ist ein wichtiges Element im globalen Klimasystem. Die genaue Bestimmung
vergangener Meeresoberflächentemperaturen ist deswegen unabdingbar für die Erforschung
der globalen Klimaprozesse und zur Evaluierung numerischer Modelle für die
Klimaprojektion. Meeresoberflächentemperaturen zu Zeiten vor der instrumentellen Messung
können anhand von klimasensitiven Biomarkern wie Alkenone und Glycerol Dialkyl Glycerol
Tetraether (GDGTs) rekonstruiert werden. In der vorliegenden Arbeit nutze ich sowohl die
obersten Abschnitte von Sedimentkernen also auch lange Kolbenlote aus dem bisher relativ
wenig untersuchten subantarktischen Südostpazifik, um das Potential der oben genannten
organischen Biomarker als Proxies für die Meeresoberflächentemperatur zu erfassen. Die
Beurteilung dieser Proxies basiert auf der Korrelation der Werte der Proxy-Indizes der
Sedimentoberflächen mit aktuellen, klimatologischen Meeresoberflächentemperaturen. Des
weiteren werden die rekonstruierten Meeresoberflächentemperaturen mit anderen regionalen
Rekonstruktionen der Meeresoberflächentemperaturen verglichen.
In den Sedimentoberflächen des Südpazifiks ist die Beziehung zwischen den Werten
der
(UK37
Alkenon-Indizes
UK’37)
und
und
den
klimatologischen
Meeresoberflächentemperaturen des World Ocean Atlas 2009 (WOA09) im niedrigen
Temperaturbereich (1 –12°C) linear. Die gleichermaßen hohen r2-Werten für beide Indizes (>
0.93)
deuten
an,
dass
die
Korrelation
beider
Alkenon-Indizes
mit
der
Meeresoberflächentemperatur auch im subantarktischen Südpazifik hoch ist und somit beide
Indizes zur Rekonstruktion der Meeresoberflächentemperaturen in den hohen südlichen
Breiten
geeignet
sind.
Trotzdem
ergeben
beide
Indizes
jedoch
unterschiedliche
Meeresoberflächentemperaturen für das Pleistozän. Das Ergebnis der UK37-basierten
Rekonstruktionen der Meeresoberflächentemperaturen entspricht dabei eher den Resultaten
anderer proxy-basierten Rekonstruktionen des Oberflächenwassers des subantarktischen
Südpazifiks, z.B. den auf Diatomeen und Foraminiferen-Vergesellschaftungen basierenden
Meeresoberflächentemperaturen und Foraminiferen-18O. Daher erscheint der UK37-Index, im
Gegensatz zum häufig genutzten UK’37-Index, besser geeignet zu sein um plausible
Rekonstruktionen der Meeresoberflächentemperaturen zu liefern.
Die Korrelation zwischen den Werten der GDGT-Indizes (TEX86 and TEX86L) in den
subpolaren und polaren Sedimentkernoberflächen (Temperaturen in den darüberliegenden
Wasseroberflächen
-2
bis
17°C)
und
den
WOA09-basierten
Jahresmitteln
der
III Meeresoberflächentemperaturen ist nicht hoch (r2 values < 0.3). Das Auftragen dieser Indizes
sowohl gegen saisonale Meeresoberflächentemperaturen als auch gegen Wassertemperaturen
an den verschiedenen hydrographischen Grenzschichten verbessert die Korrelation nicht. Die
Kombination dieser Daten mit bereits publizierten Daten von Kernoberflächen (n = 630),
welche Meeresoberflächentemperaturen von -2 bis 30°C umfassen, ergibt r2-Werte von ~ 0.8,
mit einer besseren Korrelation für TEX86 als für TEX86L. Da die Abschätzung der
Meeresoberflächentemperaturen mit Hilfe der TEX86 Kalibrierung besser mit den WOA09Meeresoberflächentemperaturen polarer und subpolarer Regionen übereinstimmt, ist dieser
Index eher zur Abschätzung der Meeresoberflächentemperaturen in diesen Regionen geeignet.
In
einem
Sedimentkern
im
Meeresoberflächentemperaturen
Südostpazifik
als
der
ergibt
Alkenon-Index
der
und
GDGT-Index
die
kältere
GDGT-basierten
interglazialen Meeresoberflächentemperaturen sind ~ 3°C kälter als die WOA09Jahresmitteltemperaturen. Da die Archaea keine saisonalen Schwankungen in der Abundanz
aufweisen, spiegeln die zu kalten, GDGT-basierten Abschätzungen wahrscheinlich nicht die
Temperaturen direkt an der Meeresoberfläche sondern die der darunter liegenden
Wasserschichten wider. Die drei Rekonstruktionen der Meeresoberflächentemperaturen in der vorliegenden
Arbeit tragen zum besseren Verständnis der ozeanographischen Oberflächenvariabilität im
subtropischen und subantarktischen Südpazifik der letzten 700.000 Jahre bei. Die Abnahme
der glazialen Temperaturen nimmt mit den Breitengraden zu, bis zu ~ 8°C in subantarktischen
und bis zu ~ 4°C in subtropischen Regionen. Die starke Abkühlung in den hohen Breiten führt
zu größeren latitudinalen Temperaturgradienten während der Kaltzeiten (Glazialen) als
während der Warmzeiten (Interglazialen). Darüber hinaus deuten die Alkenon-basierten
Meeresoberflächentemperaturen entlang des Humboldt Stroms eine massive, äquatorwärts
gerichtete Verlagerung des Frontensystems des Südlichen Ozeans sowie einen zunehmenden,
äquatorwärts gerichteten Transport von subantarktischem Wasser an. Diese Veränderungen
werden wahrscheinlich durch einen verstärkten Humboldt Strom während der Kaltzeiten
bedingt. Außerdem zeigen die GDGT-basierten Temperaturen eine verstärkte Erwärmung im
subtropischen Südpazifik während der Marinen Isotopen Stadien 11 und 13 an, welche
womöglich durch eine Umstrukturierung der Wassersäule, analog zu den modernen El-NinoVerhältnissen, bedingt ist. IV Acknowledgments
Completing this PhD work was no bed of roses. The thesis would definitely not come
into existence if not for the help from many people around me. In line with my firm belief that
science is about people and also the fact that I am a tad on the melodramatic side, here I
would like to dedicate some pages to thank these people specifically.
First of all, I would like to thank my PhD committee, comprised of Ralf Tiedemann,
Frank Lamy and Gesine Mollenhauer, for giving me the opportunity to carry out this
interesting interdisciplinary research, for their free reign-style supervision that gave me plenty
of freedom to plan and carry out my research as independently as possible, and for always
having time for me when their help was needed. I would like to thank Ralf for providing
various travel funds and for his effort in getting me the 2-months contract prolongation which
bought me some extra time to complete my thesis. Thanks to Frank my PhD study was not 3
monotonous years of laboratory work and writing. I had my fair share of adventures, i.e., seagoing experience and scientific-stays abroad, and the occasional BBQ and beers on sunny
days in the Fishtown. Heaps of thanks go to Gesine: for being the role model I could look up
to, for teaching by demonstration the importance of communication skills and scientific
integrity, for encouraging me when I hit a rough patch and for steering me in the right
direction when I could not find my way in the confusing “jungle of academia”.
I cannot thank B. David A. Naafs and Susanne Fietz enough. David has been a huge
influence in my “formative years” in academia (much like Obi-wan Kenobi to Luke
Skywalker), as he often said to me: ‘My young padawan, you have so much to learn’. I
especially appreciate his very “Dutch way” of helping – dishing out blunt, sometimes
unbearable, ego-hurting critical comments, in line with his maxim – you learn from criticism
not from praises. Thanks for being the “board” where I could bounce ideas off, for helping me
with software problems, for exchanging views on how academia should and should not be
(illustrated with stories and gossips), for barging into my office every 2 hours to shove his
latest super-nice-data-sets-that-match-his-geofantasy in my face, for motivating and bearing
with me in general. Similarly, Susanne provided stimulating discussions (on the Manolitos
a.k.a. GDGTs among many other subjects), and tried by constant repetition to get it through
my thick skull that my data are not crap. Many thanks for those fun-, gossip-, wine-, tapas-,
advices-, heated discussion-filled months in Barcelona! Thanks for having faith in my
scientific ability – it’s a source of motivation.
V Thanks are also due to various people who contributed ideas, suggestions, advices and
encouragements, which had not only improved the quality of my PhD work tremendously but
also made the journey more fun: Mahyar Mohtadi, Alfredo Martinez-Garcia, Carme Huguet,
Oliver Esper, Jens Hefter, Kalle Baumann, Thom Laepple, Peppe Cortese, Amelia Shevenell,
Valerie Masson-Delmotte, Gerald Langer, Tom Russon, Birgit Schneider and Ralph
Schneider. I especially appreciate the fact that the geographical distance between some of us
did not deter these people to lend a helping hand by putting in extra effort via numerous email
exchanges and phone / Skype calls.
Special thanks to Peppe Cortese for meticulous proofreading and improving the
quality of this thesis. Thanks also go to Susanne Fietz and Michael Shreck for their assistance
in the translation of the thesis abstract into German.
I would like to thank Rainer Gersonde, Rüdiger Stein, and Dierk Hebbeln for
providing sediment samples, without which the project would have not been possible.
Technical assistance from Walter Luttmer, Ralph Kreutz and Stephan Heckendorff (on ODV)
is also appreciated. I am grateful to Antoni Rosell-Melé and Carina B. Lange for hosting me
during the short-stays at the ICTA, Autonomous University of Barcelona and the COPAS,
University of Concepcion, respectively. I would like to thank POLMAR Graduate School for
funding my study at AWI-Bremerhaven, without which none of this work would have
happened. Administrative assistance from Claudia Sprengel and Claudia Hanfland is highly
appreciated. A SCAR Fellowship 2010/2011 is acknowledged for funding the short-stay at the
ICTA, Barcelona.
Research was a big part of my life in Bremerhaven, thus I might not have made it to
the end if not for the companionship of work-related people, who have added colors to my life
here in the past 3 years, especially Verena, Mieke, Reza, Marie, Franziska, Micha, Thomas,
Johannes, Ines and Evgenia. Thanks also go to the friendly people at ICTA, i.e., Gemma,
Bastian, Anna and especially Carme for her many hilarious and sarcastic remarks regarding
TEX86 that never fail to crack me up, and for her contagious enthusiasm for science. Special
thanks go to Mariem (my twin sister!) for being the best companion one could have during
this PhD journey.
And of course, I am indebted to my parents for their unconditional support and for
encouraging me to pursue my passion for science, even if it would mean that they have to be
thousands of miles away from me and get to see me twice in the past 3 years. Last but not
VI least, I cannot thank Raúl enough, who was always there (and thus contributed significantly to
Deutsche Bahn via his many 4-hours journeys to Bremerhaven), who bore the brunt of my
frustrations and listened to my incessant complaints with endless patience. Thanks for
distracting me from work, especially during those stressful and disillusioned dark months
when it was almost impossible to find the strength and reason to go on.
VII Table of Contents
Abstract
Zusammenfassung
Acknowledgments
Chapter 1: General overview
1.1
Global climate and sea surface temperature (SST)
1.2
Obtaining SST data
1.2.1
Instrumental SST measurements
1.2.2
Indirect SST data via proxy
1.2.2.1 Alkenone-based paleothermometry
1.2.2.2 Glycerol dialkyl glycerol tetraether (GDGT) –based paleothermometry
1.3
Subantarctic Southeast Pacific SST: present and past
1.4
Outstanding research issues
1.4.1
Occurrence of alkenones and GDGTs in the South Pacific
1.4.2
Is TEX86 paleothermometry applicable in (sub)polar regions?
1.4.3
Which alkenone index is more suitable for paleo SST reconstruction in the
South Pacific?
1.4.4
How has the Southeast Pacific SST evolved during the Pleistocene?
1.4.5
How do the TEX86-derived SSTs compare to the alkenone-derived SSTs
during the Pleistocene?
1.5
Material and methods
1.5.1
Marine sediments
1.5.2
Sample preparation
1.5.3
Alkenone analysis
1.5.4
GDGT analysis
1.5.5
Choice of environmental data for calibration
1.6
Thesis outline
I
III
V
1
2
2
3
5
7
9
12
12
12
13
14
15
15
15
17
18
20
20
21
Chapter 2: Appraisal of the TEX86 and TEX86L thermometries in the subpolar and polar
regions: implications for global core-top calibrations
2.0
Abstract
23
2.1
Introduction
24
2.2
Materials and Methods
26
2.2.1
Sediment Extraction
26
2.2.2
GDGTs Analysis
27
2.2.3
TEX86 calculation and SST estimations
29
2.2.4
Environmental data
29
2.3
Results
30
2.3.1
Relationship between individual GDGTs
30
2.3.2
Fractional abundance of GDGTs
31
2.3.3
TEX86 and TEX86L values
31
2.3.3.1 Correlation to temperature at various water depths
33
2.3.3.2 Correlation to SST of various seasons
34
2.3.4
Residuals of SST estimates
34
2.4
Discussion
38
2.4.1
GDGT contribution from methanogenic archaea
38
2.4.2
Correlation of TEX86 / TEX86L indices to subsurface temperature
39
2.4.3
Deviant SST estimates in the Arctic
40
2.4.4
2.4.5
2.4.6
2.5
Effect of sedimentation setting and/or seasonality
Implications for global core-top calibrations
Applicability in (sub)polar regions and conclusions
Acknowledgments
41
43
45
46
Chapter 3: Sea surface temperature variability in the Pacific sector of the Southern Ocean
over the past 700 kyr
3.0
Abstract
47
3.1
Introduction
48
3.2
Oceanographic setting
50
3.3
Materials and methods
51
3.3.1
Materials
51
18
3.3.2
52
 O measurement on foraminifera
3.3.3
Alkenone analysis
53
3.3.4
Alkenone-based indices and calibrations
54
3.3.5
SST gradient calculation
55
3.4
Results
55
3.4.1
Stratigraphy
55
3.4.2
South Pacific core-top alkenone calibrations
58
18
3.4.3
58
Downcore SST estimates and planktic  O values
3.4.3.1 Core GeoB 3388-1
58
3.4.3.2 Core GeoB 3327-5
58
3.4.3.3 Core PS75/034-2
59
3.5
Discussion
60
3.5.1
Alkenone-based calibrations for application in the subantarctic Pacific
60
K
K’
3.5.2
Assessing contrasting temporal trends in U 37- and U 37-derived SST
62
records
3.5.3
Southern Ocean SST evolution: circum-Antarctic comparison
64
3.5.4
Meridional SST gradients: equatorward cold water transport
67
3.6
Conclusions
70
3.7
Acknowledgments
71
Chapter 4: Diverging Pleistocene SST trends revealed by UK37 and TEX86H in the Southeast
Pacific
4.0
Abstract
72
4.1
Introduction
73
4.2
Study site
74
4.3
Materials and methods
75
4.4
Results and discussion
77
78
4.4.1
High BIT values: potential terrestrial bias on TEX86H SST?
4.4.2
Comparison with alkenone-derived SSTs
79
4.4.3
Paleoceanographic implications
82
4.5
Conclusions
84
4.6
Acknowledgments
84
4.7
Supplementary information
84
Chapter 5: Conclusions and Outlook
5.1
Summary and conclusions
5.1.1
Lipids occurrence: Where the proxies may be applied
5.1.2
Proxy indices and calibrations: Diverging views from the core-top and
downcore application
86
86
86
5.1.3
5.2
5.2.1
5.2.2
5.2.3
5.2.4
Paleo SST estimates: Pleistocene climatic implications
Perspectives for future work
Refining UK37 calibration
Refining GDGT-based SST calibrations
Constraining paleo SST reconstruction: Multi-proxy approach
Towards the “big picture”
References
Data handling
Appendix 1: Abbreviations and acronyms
Appendix 2: Units and chemical nomenclatures
Appendix 3: List of figures
Appendix 4: List of tables
Appendix 5: Additional first-authored manuscripts
88
89
90
92
95
96
97
113
114
116
117
119
120
Chapter 1 Chapter 1: General overview
1.1. Global climate and sea surface temperature (SST)
One of the pressing issues that plague humanity is climate change. A concerted effort
from the scientific community shows that the earth surface temperature has risen by almost
1°C since the Industrial Revolution [IPCC, 2007, http://www.ipcc.ch]. While there is no
doubt that climate strongly affects our societies, there is still ongoing debates on whether the
warming is attributed to natural variability or human activities. It is also crucial to know how
the climate may change in the future in order to develop the best possible mitigation policies.
Consequently, substantial amount of effort and funding have been invested in the
development of earth climate models over the years.
These earth system models are built based on our knowledge on the interaction of
various climatic components and feedback mechanisms (Figure 1.1). In this regard, the ocean
is a vital component of the global climate, through its close interaction with the atmosphere,
Figure
1.1:
Major
components
of
the
climate
system.
(from
Roger
Pielke
Sr.’s
www.pielkeclimatesci.wordpress.com).
1 Chapter 1 the land and the cryosphere. Due to its sheer size and high heat capacity, the ocean stores a
large amount of energy, which is communicated to the atmosphere via turbulent and radiative
energy exchange at the air-sea interface [Deser et al., 2010]. For instance, the condensation of
moisture from the ocean releases latent heat that contributes to drive the atmospheric
circulation, which in turn affects the sea surface temperature (SST) [Tomczak and Godfrey,
2003]. Therefore, the temperature at the sea surface is an important parameter that links the
atmosphere and the ocean, rendering it an indispensable boundary parameter in earth climate
models (Figure 1.2). Records of past SST changes are therefore useful for two reasons.
Firstly, these records help us to understand how and why the climate has changed in the past.
Secondly, we need these records to validate the climate models in order to improve the
credibility of the projections they simulate.
Figure 1.2: A schematic of the framework for projecting future climate using climate models (from Goosse et
al.’s online textbook at www.climate.be/textbook/).
1.2. Obtaining SST data
1.2.1. Instrumental SST measurements
In modern oceanography, SST data are obtained using instruments such as thermistors
and thermometers employed on ships and buoys, or the more recently developed infrared
radiometers (such as the Advanced Very High Resolution Radiometer AVHRR) mounted on
satellites. Ship-based data have lower and uneven spatial coverage but extend further back in
2 Chapter 1 time to the late 18th century. The voyage of the HMS Challenger in year 1872 marked the first
global-scale study of the ocean [Roemmich et al., 2012]. Many more organized efforts would
follow, including the International Geophysical Year (1957-1958), the World Ocean
Circulation Experiment (WOCE; 1990-2000) and the ARGO program (2004-2010), all of
which contributed important data sets of various oceanic parameters, including SST. On the
other hand, satellite SST data have global coverage at high resolution (0.25 degree), but only
go back to the early 1980s [for more details see Deser et al., 2010]. To get the best spatial and
temporal coverage to meet the demands of oceanographic research, in-situ and satellite data
could be combined into a single data set, such as the International Comprehensive
Atmosphere-Ocean Data Set (ICOADS). Alternatively, observations could be analyzed
objectively, interpolated and weighted (to account for irregular sampling in time and space) to
construct climatologies, such as the World Ocean Database (WOD). Climatological data are
the mean values (of a given oceanic parameter) over many years [Talley et al., 2011]. They
are usually constructed for the annual mean, individual months and seasons. Spatially gridded
climatological atlases, such as the World Ocean Atlas (WOA) are useful for visualization of
climatological fields and are widely used in SST proxy calibration work (Section 1.5.5).
1.2.2. Indirect SST data via proxy
Instrumental SST records span approximately the past two centuries (Section 1.2.1).
Beyond that, we have to rely on indirect SST data inferred from various proxies, which are
mostly biogeochemical in nature. Organic or inorganic fossil remains of marine organisms
represent SST proxies that are incorporated in marine sediments. These proxies are based on
observed relationships between these compounds/species and SST, e.g., the ratio of chemical
elements found in the tests of marine organisms, the relative abundance of marine organism
and the relative distribution of organic compounds within the cells. Considering the
importance of SST in the climate system (Section 1.1), since the 1950s large arrays of SST
proxies have been developed to reconstruct SST in the geological past. In a broad sense, paleo
SST proxies could be categorized into geochemical proxies (oxygen isotope, Mg/Ca, UK’37
and TEX86) and faunal/floral census proxies (foraminifera, diatom, radiolaria and
coccolithophores). Each proxy has its own advantages and weaknesses. In a recent
international Multiproxy Approach for the Reconstruction of the Glacial Ocean surface
3 Chapter 1 (MARGO) project, Kucera et al. [2005] concluded that no single proxy is “right” and “right
everywhere”. A short description of paleo SST proxies is available in Table 1.1.
Table 1.1: Overview of paleo sea surface temperature (SST) proxies.
Proxy
Faunal transfer functions
(foraminifera, radiolaria, diatom,
coccolithophores)
Working assumptions
Temperature as the major
environmental parameter causing the
changes in faunal assemblages;
relationship quantified using statistical
methods e.g., IKM [Imbrie and Kipp,
1977], MAT [Prell, 1985], GAM [Hastie
and Tibshirani, 1990], RAM
[Waelbroeck et al., 1998], ANN
[Malmgren et al., 2001], SIMMAX
[Pflaumann et al., 1996]
Limitations / Caveats
- Lack of diversity at
temperature extremes
- Evolutionary events
- Preferential preservation
References
Gersonde et al.
[2005]
Barrows and
Juggins [2005]
Cortese and
Abelmann [2002]
Saavedra-Pellitero
et al. [2011]
Oxygen isotopes  O
(foraminifera, corals, opal)
Thermodynamic fractionation between
16
18
O and O that occurs during
precipitation is a logarithmic function of
temperature
Urey [1947]
Epstein et al. [1953]
Emiliani [1955]
Hays et al. [1976]
Gagan et al. [2000]
Shemesh et al.
[1992]
Mg/Ca (foraminifera)
Incorporation of Mg in calcite varies
exponentially as a function of
temperature
Sr/Ca (corals)
Incorporation of Sr in calcite varies as
a negative function of temperature
- Vital effects (e.g.,
carbonate ion
concentration, ontogenic
effect)
18
- Unknown  O of
regional seawater
- Diagenesis (e.g., partial
shell dissolution)
- Dissolution
- Salinity
- Vital effects (e.g.,
species, pH)
- Vital effects (e.g., growth
rate and symbiont activity)
- Seawater Sr/Ca overprint
- Lateral advection
- Seasonality of production
- Non-thermal
physiological effects
18
K
U 37 (Unsaturation index of
Ketones with 37 carbons)
TEX86 (TetraEther indeX with 86
carbons)
LDI (Long chain Diol Index)
Unsaturation extent in C37 alkenones
varies as a linear function of
haptophyte growth temperature for
achieving optimum density or
enzymatic pathway
Ring moieties in glycerol dialkyl
glycerol tetraether (GDGT) archaeal
lipid vary as a linear function of growth
temperature to regulate the fluidity of
membrane
Fractional abundances of long chain
diols vary as a function of temperature
- Terrigenous overprint
- Unknown depth origin
- Uncertain source
organism
Chave [1954]
Nuernberg [1995]
Barker et al. [2005]
Beck et al. [1992]
Hendy et al. [2002]
Brassell et al. [1986]
Prahl et al. [1988]
Herbert [2003]
Schouten et al.
[2002]
Kim et al. [2010]
- Unknown mechanism of
Versteegh et al.
temperature control
[1997]
- Unknown source
Rampen et al.
organism
[2012]
- Relatively new, thus
Naafs et al. [2012]
remains to be tested
Abbreviations: IKM = Imbrie and Kipp method; MAT = Modern analog technique; GAM = Generalized additive model; RAM =
Revised analog method; ANN = Artificial neural network; SIMMAX is an acronym for a modern analog technique using a
similarity index
Over the years, advancement in analytical instrumentation and methods has enabled
more accurate quantification of compounds / elements used in marine sediment-based SST
proxies. This development has encouraged the paleoclimatic community to turn to
geochemical proxies which are generally deemed more quantitative and less time-consuming.
Among them, organic proxies i.e., UK37 (Section 1.2.2.1) and TEX86 (Section 1.2.2.2), based
on specific lipid biomarkers, are currently standard tools for inferring paleo SSTs (along with
4 Chapter 1 Mg/Ca on foraminiferal calcite). The advantage of these lipid-based proxies is that they can
be obtained simultaneously via routine organic geochemical methods, rendering multi-proxy
comparison less labor-intensive. Furthermore, these lipids, especially the GDGTs, are
ubiquitous in the global ocean spanning broad biogeographical provinces, enabling
comparison of SST records derived from a single proxy covering a large latitudinal range.
This is especially valuable in light of the discrepancies often observed in different SST
proxies due to seasonality, depth habitat of source organisms and statistical methods /
calibrations employed.
1.2.2.1. Alkenone-based paleothermometry
Alkenone paleothermometry [Brassell et al., 1986; Prahl and Wakeham, 1987] is one
of the most commonly used SST proxies nowadays and has been extensively studied in the
past two decades (for detailed overview see Appendix 5 and Herbert [2003]). It is based on
long-chain alkenones with 37 carbon atoms (Figure 1.3), found in marine haptophyte algae
such as coccolithophores of the family Noelaerhabdaceae, including the extant species
Emiliani huxleyi and Gephyrocapsa oceanica. This SST proxy is based on the alkenone
unsaturation extent, which varies as a function of growth temperature of the source organism.
In the early days of alkenone paleothermometry, it was hypothesized that the alkenones are
membrane-bound lipids [Prahl et al., 1988], and that the dependency of lipid unsaturation on
Figure 1.3: Molecular structure and standard IUPAC names of C37 alkenones with 2, 3, and 4 double bonds
(Figure from Marlowe et al. [1990], following structure verification by Rechka and Maxwell [1988] and doublebond position identification by de Leeuw et al. [1980]).
5 Chapter 1 temperature is for maintaining the membrane fluidity at low temperatures, a phenomenon
known as homeoviscous adaptation [Hazel, 1988]. However, more recent studies suggested
that alkenones are not membrane-bound [Conte and Eglinton, 1993; Sawada and Shiraiwa,
2004] and they probably serve as metabolic storage lipids [Bell and Pond, 1996; Epstein et
al., 2001]. The temperature dependence of alkenone unsaturation in the haptophyte is
therefore not straightforward and could plausibly be attributed to differences in melting point,
density or enzymatic optima of biochemical pathways of the alkenones [Epstein et al., 2001].
In spite of our limited understanding on the biochemical role of the unsaturation in the
alkenones, the empirical correlation between the unsaturation extent and temperature has
resulted in numerous high-quality paleo SST records spanning various geological time-scales.
Brassell et al. [1986] first proposed an index known as UK37 (U = unsaturation, K = Ketone
(alkenone), 37 = chain length of ketone) to quantify the unsaturation extent in alkenones:
K

U 37
C 37:2  C 37:4
C 37:2  C 37:3  C 37:4
As pointed out by Bendle and Rosell-Melé [2004], there is no apparent biogeochemical
rationale for the form of the numerator in this index and it was derived by empirical trial-anderror (until an index that correlates best with temperature is obtained). Indeed, Calvo et al.
[2002] speculated that the inclusion of C37:4 in the numerator might lead to “overly” low index
values, hence an overestimation of cooling. This index was later simplified as UK’37 by Prahl
and Wakeham [1987] by removing the C37:4 alkenones from the equation since the inclusion
of this compound did not improve the unsaturation extent – temperature correlation in
laboratory E. huxleyi cultures.
K'
U 37

C 37:2
C 37:2  C 37:3
Subsequent studies, mostly in the mid and low latitudes, found no C37:4 alkenones in the
marine sediments and suspended matters in sea water above SST of 15°C. As a result, most of
the alkenone-related studies available in literature at present are based on the simplified UK’37
index.
The most commonly used calibration to convert the alkenone index values into SSTs
is Prahl et al. [1988]’s culture-based UK’37 calibration. In this work, Prahl and co-workers
grew E. huxleyi cultures (strain VAN55 isolated from North Pacific waters) at temperatures
6 Chapter 1 between 8 and 25°C, and observed linear relationships between growth temperatures with
UK37 and UK’37 values (the latter showed a slightly better correlation). The UK’37 calibration
has since been attested by two global core-top UK’37 calibrations, proposed by Müller et al.
[1998] (n=370) and Conte et al. [2006] (n=592), respectively. All three of these calibrations
are statistically similar.
1.2.2.2. Glycerol dialkyl glycerol tetraether (GDGT) –based paleothermometry
It was not until year 2002, 16 years after the introduction of alkenone
paleothermometry, that another organic SST proxy came along. Schouten and co-workers
[2002] found that the relative distribution of GDGTs (with different ring moieties; Figure
1.4) in 42 marine sediment samples varies as a function of SST. The authors speculated that
the source organisms of these lipids, i.e., non-thermophilic Thaumarchaeota (formerly known
as Marine Group 1 Crenarchaeota), increase the ring moieties in their membrane lipids at
higher temperatures. Such modification of archaeal lipids as a response to temperature has
been observed previously in hyperthermophilic Crenarchaeota [De Rosa and Gambacorta,
1988; Gliozzi et al., 1983; Uda et al., 2001]. The increased ring moieties were thought to
provide a better packing of the membrane at higher temperatures [Gliozzi et al., 2002]. As
mentioned by Schouten et al. [2002], microbiologists [De Rosa and Gambacorta, 1988;
Gliozzi et al., 1983; Uda et al., 2001] quantified the ring moiety – temperature relationship in
extremophiles using the weighted average number of rings in GDGTs. However, the ring
index values in marine sediments did not correlate well with overlying SSTs. After some trialand-errors, Schouten et al. [2002] found that the best fit was obtained by an index termed
TEX86 (TetraEther index of tetraethers consisting of 86 carbon atoms):
TEX 86 
GDGT - 2  GDGT - 3  Cren'
GDGT - 1  GDGT - 2  GDGT - 3  Cren'
The two most abundant isomers, i.e., GDGT-0 and Crenarchaeol, were not included in the
index because the occurrence of GDGT-0 is not exclusive to Thaumarchaeota – they were
previously found in methanogenic archaea [Schouten et al., 2000]. Secondly, the abundance
of Crenarchaeol is one magnitude higher than GDGT-1, -2 and-3, thus it might overwhelm the
correlation found between these latter compounds with temperature.
7 Chapter 1 Figure1.4: Molecular structure of isoprenoid glycerol dialkyl glycerol tetraether (GDGT) lipids with different
ring moieties (from Ho et al. [2011]).
Later studies [Kim et al., 2010] with a much larger data set (n= 396) showed that a
better correlation was found between SST and a modified GDGT-based index known as
TEX86L:
L
 Log(
TEX 86
GDGT - 2
)
GDGT - 1  GDGT - 2  GDGT - 3
Notably, this index emerged as the best among all 1953 combinations consisting of the 6
archaeal isoprenoid GDGTs (Figure 1.4) assessed by Kim and co-workers. Nevertheless, as is
the case with alkenone paleothermometry, these GDGT indices are empirical. While there is a
biochemical reason for the temperature dependency of the ring moieties in GDGTs, the
definition of the indices are derived based on the outcome of many trial-and-errors, performed
with a priority in finding the one that fits best to SST.
Calibrations commonly used to convert the index values into SST estimates are based
on marine sediment studies [Kim et al., 2008; Kim et al., 2010; Schouten et al., 2002]. Marine
8 Chapter 1 Thaumarchaeota have not yet been successfully cultured. Whilst a mesocosm study on North
Sea waters by Wuchter et al. [2004] demonstrated a first-order relationship between TEX86
and growth temperature, the equation is different from that found in marine sediments, casting
doubt on its applicability in paleo SST reconstruction. Interestingly, recent development in
TEX86 paleothermometry has seen several cases in which the index, thereby also the
calibration, was adapted for a specific study site in order to obtain “more reasonable” SST
estimates (shortly described in Table 1.2). The global applicability of these indices is not
granted as they have never been applied elsewhere except in the study they were proposed.
Table 1.2: Short description of several modified GDGT-based indices and calibrations that deviate from the
commonly applied global core-top calibrations of Schouten et al. [2002], Kim et al. [2008] and Kim et al. [2010].
Index
Calibration
Studied area and
Reference
timescale
TEX 86 ' 
GDGT - 2  GDGT - 3  Cren'
GDGT - 1  GDGT - 2  Cren'
104 unspecified core-top data
Arctic; Paleocene –
Sluijs et al.
vs. atlas SST
Eocene thermal
[2006]
maximum
1
GDGT - 1

1
TEX86 GDGT - 2  GDGT - 3  Cren'
Kim et al. [2008]’s TEX86 data
Global; Eocene-
Liu et al.
vs. atlas SST data (n=287)
Oligocene climate
[2009]
Kim et al. [2008]’s core-top
Antarctic Peninsula
Shevenell et
TEX86 data (n=287) vs. atlas
(AP), Holocene
al. [2011]
transition
TEX86 (see text above, proposed by Schouten et
al., [2002])
SST; combined with 7 AP
core-top TEX86 vs. measured
SST
1.3. Subantarctic Southeast Pacific SST: present and past
To get a grasp on how the climate system operates, one first needs to understand its
components, including the ocean (Section 1.1). Oceanographic studies could be divided into
two domains, i.e., modern oceanography and paleoceanography. Historical SST data (e.g.,
from merchant ship-based measurements) are heavily concentrated in the North Atlantic,
western South Atlantic and northern Indian oceans [Deser et al., 2010]. The Southern Ocean,
on the other hand, is poorly sampled due to a scarcity of trading ports in the Southern
hemisphere, expensive logistics, and its uninviting rough seas. The same could be said about
the paleo domain – progress in Southern Ocean research, especially in the Pacific sector, is
hampered by a dearth of data. This situation partly arises from logistical difficulty since most
well-funded oceanographic institutions are located in the Northern hemisphere on both sides
of the North Atlantic. Another contributing factor is the “obsession” with the North Atlantic
9 Chapter 1 since the early days of paleoceanography. As dramatically put by Huybers and Wunsch
[2010], “Like the Genesis story, the idea that the North Atlantic Ocean meridional
overturning circulation is the major controller of the climate system has taken on an almost
mythic status”. Nevertheless, recent studies suggest that the upwelling of intermediate and
deep waters in the Southern Ocean is one of the main components of the global overturning
circulation, and it influences the exchange of heat and carbon between the deep and the
surface ocean and the atmosphere [Marshall and Speer, 2012] Therefore, not unlike the North
Atlantic (one of the locations in the global ocean where deep waters are formed), the Southern
Ocean too plays an important role in global climate [see review by Fischer et al., 2010], via
its linkages with the intense wind field [Toggweiler and Russell, 2008] and its seasonally
varied sea-ice cover [Stephens and Keeling, 2000]. Incidentally, SST is a key element in these
physical processes, as it either steers them, or it is modified by them.
The geographic pattern of SST is strongly linked to, and controlled by, atmospheric
and oceanic processes. Among the latter, physical processes such as heat transport by currents
and vertical mixing play a major role. The single most important surface current in the
Southern Ocean is the formidable Antarctic Circumpolar Current (ACC) that flows
unimpeded around Antarctica, connecting the Atlantic, the Pacific and the Indian Oceans. It is
also the largest current in the world, transporting around 110 Sv of water [Cunningham et al.,
2003], and its latitudinal position and intensity have implications for the global ocean and
climate. In the east Pacific sector of the Southern Ocean, a very vigorous eastern boundary
current i.e., the Peru-Chile Current (PCC; also known as the Humboldt Current and the Peru
Current) is formed as a result of the bifurcation of the ACC as it approaches the South
American continent (Figure 3.1). The PCC brings cold and nutrient-rich subantarctic water
equatorward, feeding the east Pacific equatorial cold tongue. This current is therefore a heat
conduit and forms the link between high and low latitudes in this sector of the Southern
Ocean. Furthermore, it provides nutrients that sustain the high productivity in the upwelling
regions off northern Chile and Peru, without which the profitable fishery industry here, one of
the largest on the planet [Chavez and Messié, 2009], would perish. More detailed description
of these surface currents is available in Chapter 3.
In spite of the remarkable difficulties in logistic and in finding suitable and datable
sediments, several studies have been conducted in the Pacific sector of the subantarctic
Southern Ocean to examine past SST changes (see Figure 1.5). These studies made use of
various paleo SST proxies described in section 1.2.2 and Table 1.1. An emerging picture
10 Chapter 1 from these studies is that different proxies suggest dissimilar extents of glacial cooling in this
region, plausibly due to differences in production seasonality and habitat depth of the source
Figure 1.5: Literature review of paleo sea surface temperature reconstruction in the subantarctic South Pacific
south of 40°S. Red squares denote continuous downcore reconstructions; blue triangles denote the Last Glacial
Maximum (LGM) time-slice study of Luz [1977]; black circles denote LGM time-slice study of Gersonde et al.
[2005]. For more details see Table 1.3.
Table 1.3: Details of previous paleo sea surface temperature studies in the subantarctic Southeast Pacific (south
of 40°S).
Site
Time scale
ODP 1233
70 kyr
27 kyr
MD07-3128
E11-1
Type of Proxy
Alkenone (U
K’
37
)
Coccolithophorid TF
K’
37
E11-4
E15-4
E15-6
60 kyr
LGM (time-slice)
LGM (time-slice)
110 kyr
LGM (time-slice)
LGM (time-slice)
LGM (time-slice)
LGM (time-slice)
LGM (time-slice)
LGM (time-slice)
LGM (time-slice)
Alkenone (U )
Foraminiferal TF
Diatom TF
Mg/Ca on N. pachyderma (s)
Foraminiferal TF
Diatom TF
Foraminiferal TF
Diatom TF
Diatom TF
Diatom TF
Diatom TF
E15-12
E19-7
E20-18
LGM (time-slice)
LGM (time-slice)
LGM (time-slice)
Diatom TF
Diatom TF
Foraminiferal TF
E11-2
E11-3
E25-10
LGM (time-slice)
Foraminiferal TF
PS58/271-1
LGM (time-slice)
Diatom TF
DWBG 70
LGM (time-slice)
Foraminiferal TF
Abbreviations: LGM = Last Glacial Maximum; TF = transfer function
Glacial-interglacial
amplitude
~7°C
~5°C
~8°C
2.3°C
0.7°C
~3°C
2.5°C
0 °C
1.5°C
1.6°C
0.3°C
0.7°C
0.2°C (LGM warmer than
present)
0.9°C
1.4°C
0.2°C (LGM warmer than
present)
4.9°C
1.3°C
3.4°C
Reference
Lamy et al. [2004];
Kaiser et al. [2005]
Saavedra-Pellitero et
al. [2011]
Caniupán et al. [2011]
Luz [1977]
Gersonde et al. [2005]
Mashiotta et al. [1999]
Luz [1977]
Gersonde et al. [2005]
Luz [1977]
Gersonde et al. [2005]
Gersonde et al. [2005]
Gersonde et al. [2005]
Gersonde et al. [2005]
Gersonde et al. [2005]
Gersonde et al. [2005]
Luz [1977] Luz [1977] Gersonde et al. [2005]
Luz [1977] 11 Chapter 1 organisms. SST estimates based on diatom assemblages (max. 1.5°C; Gersonde et al. [2005])
and Mg/Ca on planktonic foraminifera N. pachyderma (~3°C; Mashiotta et al. [1999])
suggested less severe glacial cooling, as opposed to stronger cooling indicated by alkenone
unsaturation (UK’37) (~7°C; Caniupán et al. [2011]; Kaiser et al. [2005]; Lamy et al. [2004]),
coccolithophorid assemblages (6°C; Saavedra-Pellitero et al. [2011]), and foraminiferal
assemblages (max. 5°C; Luz [1977]). Since these studies linked the SST changes to the
latitudinal migrations of key climatic elements in the Southern Ocean - the ACC and its
associated oceanic frontal systems, the southern hemisphere Westerlies and the sea-ice extent
– dissimilar glacial cooling leads to diverging view on the latitudinal positions of these
climatic features.
1.4. Outstanding research issues
1.4.1. Occurrence of alkenones and GDGTs in the South Pacific
To date there is no report on the occurrence of alkenones and GDGTs in surface
sediments between the shallow waters off New Zealand [Sikes et al., 1997] and the
continental shelves off Chile [Kim et al., 2002], casting some uncertainties on the potential of
these biomarkers for paleo SST reconstruction work in this region.
1.4.2. Is TEX86 paleothermometry applicable in (sub)polar regions?
(b) TEX86L (a) TEX86 r2=0.77; n=396
r2=0.86; n=396
Figure 1.6: Crossplots of GDGT-based indices (a) TEX86 and (b) TEX86L with satellite SSTs from Kim et al.
[2010]. Blue symbols represent subpolar and polar samples from the Arctic, the Barents Sea, off Svalbard, the
Southern Ocean, and off Antarctica (see Figure 2.1 for location). Purple whisker bars denote residual standard
error (±1σ).
The applicability of TEX86 in the subpolar and polar regions has not been rigorously
tested, especially in the context of late Pleistocene glacial-interglacial SST shifts. A recent
12 Chapter 1 reconstruction by Shevenell et al. [2011] on Holocene SST changes off Antarctic Peninsula
demonstrated that although the TEX86 index did show climatic shift that were qualitatively
consistent with other proxy data, both the latest global core-top calibrations of TEX86 and
TEX86L of Kim et al. [2010] resulted in unrealistic SST estimates. As mentioned by Kim et al.
[2010], the crossplots of both indices versus SST (Figure 1.6) show large scatter at the low
temperature end, where the indices seem to be invariant with temperature. The fact that the
scatter at low temperature in the TEX86L-SST correlation is as large as, if not larger, than that
in the TEX86-SST relationship, casts doubts on the justification of the TEX86L as the better
index at low temperature range (recommended by Kim et al. [2010]). The marginally
improved r2 value in the TEX86L-SST relationship is probably a statistical artifact due to the
amplified variance of TEX86L at the low temperature end. Furthermore, since most of the data
forming the scatter are retrieved from continental shelves prone to terrestrial GDGT input, it
is not clear whether the scatter is due to the physiological limit of archaeal response to
temperature, or other underlying cause (e.g., terrigenous overprint).
In addition, there are very few data from the Southern Ocean and the Pacific (see
Chapter 2; Figure 2.1), so the TEX86-SST relationship in these regions is currently not well
constrained. Although a study by Ho et al. [2011] suggested that there is no apparent regional
bias in the South Pacific (see Appendix 5), the study is based on only 20 data points. More
data are necessary to appraise the applicability of TEX86 paleothermometry in this region.
A further complicating factor is represented by the two modified versions of TEX86,
namely the TEX86L and TEX86H for application at sites <15°C proposed by Kim et al. [2010].
Applying different indices at opposite temperature ends adds more uncertainty in comparing
SST records from high and low latitudes (which are not directly comparable since they are
based on different indices and calibrations). Therefore, work is needed in order to establish a
universal calibration which is applicable throughout the entire temperature range.
1.4.3. Which alkenone index is more suitable for paleo SST reconstruction in the South
Pacific?
The few marine sediment and water column studies in the high latitudes show
contradictory findings on the applicability of UK37 and UK’37. The UK37 index appears to be a
better SST proxy in the Nordic Sea [Bendle and Rosell-Melé, 2004] while the UK’37 seems to
13 Chapter 1 be a better choice in the Southern Ocean [Sikes et al., 1997]. Multi-proxy downcore SST
reconstructions in the North Atlantic by Bard [2001] showed that the UK37, instead of the
simplified UK’37, resulted in SST estimates that are more comparable with other proxy data.
This is somehow to be expected, considering the working principle of alkenone
paleothermometry, which dictates that the unsaturation extent in alkenones increases with
decreasing temperature. The abundance of the more unsaturated alkenones (i.e., C37:4)
becomes numerically important at low temperatures, hence the exclusion of this compound
from the SST index might undermine its predictive power (as is the case in the simplified
UK’37).
Notwithstanding, Sikes et al. [1997] found better correlation in the UK’37-SST
relationship in the Southern Ocean. Notably, the UK37 values in the Southern Ocean surface
sediment reported by Sikes et al. [1997] are systematically lower than those observed in the E.
huxleyi culture of Prahl et al. [1988] (The UK’37-SST relationship in the latter study has been
attested by two global core-top calibration studies and is the most commonly employed
calibration). The discrepancy between these two data sets is due to the higher relative
abundance of C37:4 alkenones in the Southern Ocean surface sediment, which amounted to
~15% at 15°C and even as high as 38% at 9°C, while the cultured E. huxleyi started producing
this compound at relatively low abundance at 10°C (absolute concentration data not reported).
Therefore, it is still debatable which alkenone index is more suitable for paleo SST
reconstruction in the Southern Ocean.
1.4.4. How has the Southeast Pacific SST evolved during the Pleistocene?
In spite of climatically important oceanic features in the east Pacific sector of the
Southern Ocean such as the ACC and the PCC, studies are scarce in this region (see Chapter
3). In the past 50 years, there have been fewer than a dozen studies on paleo SST here using
marine sediment-based proxies (see Section 1.3). These SST records also do not go very far
back in time, with the longest record [Kaiser et al., 2005] spanning merely 70 kyr. Therefore,
the late Pleistocene SST evolution prior to MIS 4 here is virtually unknown, a fact that has
important implications, as we lack information on the warmer-than-present MIS 5 and MIS
11, and on the Mid-Brunhes Event (MBE), a climatic reorganization of the glacial-interglacial
cycles which is well-expressed in the ice core records from Antarctica and is thought to play a
14 Chapter 1 key role in the global carbon cycle. The data scarcity also impedes circumpolar comparison of
climatic evolution in the Southern Ocean and Antarctica.
1.4.5. How do the TEX86-derived SSTs compare to the UK37-derived SSTs during the
Pleistocene?
As the MARGO community [Kucera et al., 2005] concluded, no single proxy is
“right” and “right everywhere”, highlighting the importance of a multi-proxy approach to
better constrain SST reconstruction. Within the context of organic SST proxies, multi-proxy
studies comparing alkenone- and GDGT-derived SST estimates are still not widespread but
are on the rise, including studies from Castañeda et al. [2010]; Huguet et al. [2006b], ,
Huguet et al. [2011], , Lopes dos Santos et al. [2010] and McClymont et al. [2012]. Most of
these studies are based on sediment records from shallow continental margins with high
sedimentation rates. The emerging picture from these studies is a complex one, with some
studies suggesting diverging SST evolutions [Huguet et al., 2006b; Lopes dos Santos et al.,
2010], while other studies do not support this scenario [Castañeda et al., 2010; Huguet et al.,
2011; McClymont et al., 2012]. In light of the recent finding that sedimentation settings
(coastal vs. offshore) result in contrasting biases in TEX86-derived SSTs [Leider et al., 2010],
a study based on material from an open ocean setting with moderate to low sedimentation rate
would further add to our understanding of these organic SST proxies.
1.5. Material and methods
Marine sediments are the climatic archive used in this thesis to address the issues
discussed above (Section 1.4). These sediment samples were analyzed using organic
geochemical methods outlined in the following sub-sections.
1.5.1. Marine sediments
Various chemical (e.g., degradation and remineralization) and physical (e.g.,
advection) processes occurring in the water column during the sinking of biomarkers from the
sea surface to the sediment might alter the empirical relationship between biomarkers and
environmental parameters, resulting in dissimilar calibrations in the suspended organic matter
15 Chapter 1 and in the sediment [e.g., Conte et al., 2006]. Hence, temporally and spatially averaged
surface sediments are appropriate for assessing the applicability of a proxy in downcore
reconstruction, as the sediments have undergone similar processes in the water column before
being incorporated into the sediments. In this thesis, an extensive set of surface sediments
were analyzed (Figure 1.7). These sediments are the upper most 0-1 cm of multicores,
retrieved during various cruises in under-sampled regions such as the North Pacific, the Arctic
and the Southern Ocean (see Section 2.6). The core sites are strategically located to fill the
geographical voids in the global core-top calibration data sets (see Figure 2.1).
Figure 1.7: Location of marine sediments used in this study. Red triangles denote long piston cores for
downcore reconstruction (Chapter 3 and 4); black circles denote surface sediments used for TEX86 calibration
(Chapter 2); blue crosses denote surface sediments used for UK37 calibration (Chapter 3).
To examine the evolution of SST in the Southeast Pacific at different latitudes on
orbital time scale, we analyzed three marine sediment cores obtained via piston coring along
the latitudinal range of the Peru-Chile Current (see Section 3.3.1). These sediments span at
least 500 kyr, and the oldest core extends back to 700 kyr, according to the stratigraphic
framework established via graphic alignment of benthic foraminiferal 18O to the global stack
LR04 [Lisiecki and Raymo, 2005], or visual tuning of the SST record to the air temperature
record in the EPICA ice core at Dome C, Antarctica (see Section 3.4.1).
16 Chapter 1 1.5.2. Sample preparation
Standard organic geochemical techniques were employed in this work. All sediment
samples were freeze-dried and homogenized before being subjected to extraction using
organic solvents. Freeze-drying is preferred for sediments used for alkenone analysis because
air-drying (commonly used for samples prior to microfossil analysis) might lead to significant
loss of alkenones and potentially bias the temperature estimation [McClymont et al., 2007].
Several extraction methods were applied in this work, such as the Accelerated Solvent
Extractor (DIONEX 200), ultrasonication, and microwave-assisted extraction (see Section
2.2.1). It is conceivable that different extraction techniques might lead to dissimilar recovery
rates of extractable compounds which could affect the absolute compound concentration per
sediment weight. However, this is not a concern for this thesis since the focus is not on
comparing the concentration of biomarkers at various study sites. Furthermore there is no
evidence of extraction method-induced bias in index values.
Figure 1.8: Types of biomarkers and their precursors in the three domains of life. Red circles denote biomarkers
used in this study for paleo SST reconstruction. (Figure courtesy of Florian Rommerkirschen; in Gaines et al.
[2008]).
17 Chapter 1 The total lipid extract is composed of a large suite of biomarkers derived from
organisms spanning archaea, bacteria and eucarya (i.e., the three domains of life), as
illustrated in Figure 1.8. To have purer samples for a better quantification, the total lipid
extracts were partitioned into different fractions by means of silica gel chromatography or via
a high performance liquid chromatography system equipped with a silicon dioxide column
(details in Section 2.2.1 and 3.3.3).
Organic compounds move through silica or alumina columns at different rates
depending on the type of column (stationary phase) and solvents (mobile phase) used, as
illustrated in Figure 1.9. There is higher affinity between silica / alumina and polar
compounds, hence the less polar alkenones elute through the columns before the more polar
GDGTs.
Figure 1.9: Schematic of silica or alumina column chromatography to separate a mixture of organic compounds
into different fractions prior to analysis. (Figure from explow.com/chromatography)
1.5.3. Alkenone analysis
After the separation of total lipid extracts, the fraction containing alkenones was
saponified using weak potassium hydroxide (KOH) in methanol to remove ester-bound
18 Chapter 1 compounds such as alkenoates which might co-elute with alkenones in gas chromatography.
In general, saponification is not a standard protocol of alkenone analysis, especially for
pelagic sediments which are usually depleted in organic matter. However, it was essential for
the sediment core-tops examined here as co-elution of an unknown compound with the C37:4
alkenones occurred in many samples from sites with overlaying SST < 10°C, resulting in
similar UK’37 values but different UK37 values (Figure 1.10). The co-elution artificially
increased the %C37:4 and lowered the UK37-derived temperature estimates. Notably, this
unknown compound was not found in deeper sediments in piston core PS75/034-2, suggesting
that it is probably not as refractory as the alkenones and was degraded in the upper few
centimeters. Saponification of six samples with %C37:4 of 10% to 20% did not result in
different temperature estimates and %C37:4 values, as they were comparable to those obtained
from untreated samples. Therefore, the alkenone samples for core PS75/034-2 were not
saponified prior to analysis.
0.8
UK'37
UK37
1:1 Line
Without Saponification
0.6
0.4
0.2
0.0
‐0.2
‐0.4
‐0.6
‐0.8
‐0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
With Saponification
Figure 1.10: South Pacific core-top alkenone index values with and without saponification.
Alkenones were analyzed using gas chromatography (GC) equipped with a flame
ionization detector. In a GC system, injected samples are vaporized and pushed through the
column by a carrier gas (mobile phase). The separation of different compounds is achieved
via differences in boiling point, molecular size and affinity for the mobile phase. The
identification of alkenones was done by comparing the sample chromatogram to that of a
reference standard extracted from an Emiliania huxleyi culture (similar compounds should
19 Chapter 1 have similar retention times). The instrumental precision for alkenones analysis is estimated
to be ~0.2 °C based on replicate measurements.
1.5.4. GDGT analysis
Unlike for alkenones, gas chromatography is not adequate for analyzing the relatively
high-molecular-weight GDGTs. Instead, a high performance liquid chromatography (HPLC)
was used. In this work, GDGT analysis was carried out using HPLC systems coupled to a
mass spectrometer, operating with an atmospheric pressure chemical ionization (APCI)
interface (see Section 2.2.2 for details). The TEX86 values obtained using these systems were
not compared directly, but significant bias is unlikely since an interlaboratory comparison
study suggests that different HPLC systems result in comparable TEX86 values [Schouten et
al., 2007]. Samples were passed through polytetrafluoroethylene (PTFE) filters prior to
injection to prevent clogging in the column. The detection of the GDGTs was achieved by
Selected ion Monitoring (SIM) of the [M+H]+ ions in the m/z ranges described by Hopmans
et al. [2000].
1.5.5. Choice of environmental data for calibration
To calibrate alkenone- and GDGT-based proxies, the indices were plotted against
various environmental data sets retrieved from the World Ocean Atlas 2009 [Locarnini et al.,
2010]. These include the annual mean and seasonal SSTs, in addition to temperature data at
different water depths. As mentioned in Section 1.2.1, several reanalyzed and climatological
SST products are available for research purposes. The WOA09 data sets were employed in
this work, in consistency with most calibration studies in the past which opted for earlier
versions of WOA data sets [e.g., Kim et al., 2008; Müller et al., 1998; Sikes et al., 1997].
However, Kim et al. [2010] used the NSIPP AVHRR Pathfinder and Erosion Global 9 km
climatology data [Casey and Cornillon, 1999] in their most recent global core-top TEX86
study. This data set is not significantly different from the WOA09 data set, rendering
calibrations based on these data sets comparable. The deviations between these data sets occur
at the low temperature end (Figure 1.11), where the Pathfinder SST values are colder than
those from the WOA09, and can get as low as -2.4°C. Such low SST values are probably
20 Chapter 1 errors related to the processing of satellite data, given the theoretical coldest seawater
temperature is -1.8°C.
35
1:1 line
Pathfinder SST (°C)
30
25
20
15
10
5
0
r2 = 0.986
‐5
‐5
0
5
10
15
20
WOA09 SST (°C)
25
30
35
Figure 1.11: Correlation between the climatology SST data sets from the NSIPP AVHRR Pathfinder and
Erosion Global 9km [Casey and Cornillon, 1999] and the World Ocean Atlas 2009 [Locarnini et al., 2010].
1.6. Thesis outline
The outstanding research issues outlined in Section 1.4 are addressed in Chapter 2 to
4 of this thesis, presented in the form of three manuscripts submitted or to be submitted to
international peer-reviewed journals. Two additional manuscripts produced during the course
of this PhD study, i.e. a review paper on alkenone paleothermometry and a Pacific TEX86
core-top calibration paper based on the data generated during MSc. study, are included in
Appendix 5.
Chapter 2 presents 160 new (sub)polar core-top TEX86 and TEX86L data that greatly
enhance the spatial coverage and the temperature range of the previous global core-top
calibration of Kim et al. [2010]. In this chapter, the implications of new data addition for the
global core-top calibrations are appraised. The suitability of TEX86 and TEX86L as SST proxy
in subpolar and polar regions are further evaluated based on temperature residuals. The
extensive compilation (~600 data) comprising data presented here and published data is
rigorously assessed for any potential bias between the high- and low latitudes, and between
different sedimentation settings.
Chapter 3 contributes to the spatial mapping of alkenones in surface sediment in the
South Pacific, which is of great importance for the planning of future paleoceanographic
21 Chapter 1 studies in this region. The assessment of the suitability of alkenone indices as SST proxy in
the Southeast Pacific is carried out with a two-pronged approach, i.e., by examining the
correlation of core-top alkenone index values with modern SST; and by comparing the
alkenone indices-derived SST records with other proxy records. After establishing the more
appropriate alkenone index for paleo SST estimation, the SST records are used to infer the
latitudinal migration of the oceanic frontal systems during the late Pleistocene, with an
emphasis on the glacial-interglacial amplitude and notable climatic events such as the MBE.
Chapter 4 presents a comparison of alkenone- and GDGT-based Pleistocene SST
records in the Southeast Pacific. Several scenarios related to climatic events and proxies are
suggested to reconcile the differences in the absolute values and the temporal trends of these
SST records.
Chapter 5 summarizes the findings discussed in abovementioned chapters. In light of
these findings, perspectives for future work to further improve the proxies and their
application are proposed.
22 Chapter 2 Chapter 2: Appraisal of the TEX86 and TEX86L thermometries in the
subpolar and polar regions: implications for global core-top calibrations
Sze Ling Ho a*, Gesine Mollenhauer a, Susanne Fietz b, Alfredo Martínez-Garcia b,1, Frank
Lamy a, Gemma Rueda b, Konstanze Schipper a,c, Marie Méheust a, Antoni Rosell-Melé b,d,
Rüdiger Stein a, Ralf Tiedemann a
a
Alfred Wegener Institute for Polar and Marine Research, P.O.Box 12 01 61, 27515 Bremerhaven, Germany.
Institut de Ciència i Tecnologia Ambientals, Universitat Autònoma de Barcelona, 08193 Bellatera, Spain.
c
Department of Geosciences, University of Bremen, 28334 Bremen, Germany.
d
Also at Institució Catalana de Recerca i Estudis Avançats, Barcelona, Catalonia, Spain.
1
Presently at Geological Institute, Swiss Federal Institute of Technology Zürich, NO G 55, Sonneggstrasse 5,
CH-8092 Zürich, Switzerland.
b
(Under review with Geochimica et Cosmochimica Acta)
2.0. Abstract
TEX86 (TetraEther indeX of tetraethers consisting of 86 carbon atoms)
paleothermometry is an organic sea surface temperature (SST) proxy based on the archaeal
isoprenoid glycerol dialkyl glycerol tetraethers (GDGTs). In this study, we improved the
spatial coverage of the global core-top TEX86 calibration by contributing additional core-top
data from the subpolar and polar regions, and appraised the applicability of the TEX86 and a
modified version, i.e., TEX 86L , in these regions. The SST estimates derived from both TEX86
and TEX 86L are anomalously warm in the Arctic especially in the vicinity of the sea ice
margin, calling for caution in interpreting paleo TEX86/ TEX 86L reconstruction in this region.
Judging from the temperature residuals, the TEX86 calibration resulted in better SST estimates
in subpolar and polar regions, especially the Southern Ocean and the North Pacific, where the
TEX 86L -inferred SSTs are considerably warmer than summer SSTs, in spite of the fact that the
TEX 86L is recommended for application at sites with SST < 15°C. Linear regressions through
our compiled global data set (excluding the Arctic data) resulted in TEX86 (n = 482) and
TEX 86L (n = 480) calibrations with r2 values of 0.8 and 0.73 respectively, further confirm the
robust relationship between both GDGT-based indices and SST for a broader geographical
coverage. Between these indices, TEX86 appears to be a more suitable SST proxy for the
23 Chapter 2 application in subpolar and polar regions, judging from better correlation in TEX86-SST
calibration and smaller TEX86-derived temperature estimates residuals in these regions.
2.1. Introduction
TEX86 (TetraEther indeX of tetraethers consisting of 86 carbon atoms)
paleothermometry is an organic sea surface temperature (SST) proxy based on the relative
distribution of isoprenoid glycerol dialkyl glycerol tetraethers (GDGTs) in archaeal lipids
[Schouten et al., 2002]. These lipids appear to be mostly biosynthesized by Thaumarchaeota
(previously known as Group I.1a Crenarchaeota) [Brochier-Armanet et al., 2008] that are
omnipresent in the global ocean. The number of cyclopentane moieties of these archaeal
GDGT lipids in marine sediments was found to demonstrate an empirical linear relationship
with the annual mean temperature of the overlaying sea surface water [Schouten et al., 2002],
suggesting its potential as a proxy for paleo-environmental conditions. Later on, a mesocosm
study confirmed that indeed the relative distribution of the GDGTs in seawater varied as a
function of the growth temperature of the source organism [Wuchter et al., 2005].
Since then, many more studies have been carried out to scrutinize the potential of
TEX86, revealing that the ring numbers in the GDGTs seem to be responsive solely to the
growth temperature, and are not influenced by salinity and nutrient availability [Wuchter et
al., 2004], sediment maturity [Schouten et al., 2004], or grazing [Huguet et al., 2006a]. It was
also found that the isoprenoid GDGTs are less susceptible to long distance lateral transport
relative to alkenones used for the more established organic SST proxy [Mollenhauer et al.,
2008; Mollenhauer et al., 2007; Shah et al., 2008]. In addition, the fact that the GDGTs are
found at all latitudes of the global ocean, including the polar regions that are often devoid of
K'
alkenones, and hence preclude the application of the well established organic SST proxy U 37
[Kim et al., 2010], suggest a potential advantage of the TEX86 paleothermometer in this
region.
However, there are some uncertainties and caveats associated with the GDGT-based
proxy. For instance, the GDGTs have been found throughout the water column and in the
sediments [Lipp and Hinrichs, 2009], sometimes even more abundant at subsurface than
surface waters [Huguet et al., 2007; Sinninghe Damsté et al., 2002a; Wuchter et al., 2005] and
may reflect subsurface water temperature in some settings [Huguet et al., 2007; Lee et al.,
24 Chapter 2 2008]. Moreover, a lack of understanding on the ecology, metabolic pathways and energy
sources of the Thaumarchaeota, and the processes that control the sedimentary deposition of
GDGTs, makes it difficult to constrain the water depth where the sedimentary GDGTs might
have originated, thus complicating the interpretation of TEX86 values as SST estimates.
Initially, TEX86 has been defined by Schouten et al. [2002] based on a global core-top
study (n = 44) as:
TEX 86 
[GDGT  2]  [GDGT  3]  [Cren' ]
[GDGT  1]  [GDGT  2]  [GDGT  3]  [Cren' ]
where GDGT-1, GDGT-2, GDGT-3 denote GDGTs containing 1, 2 and 3 cyclopentane
moieties respectively; and Cren’ the crenarchaeol regioisomer. Recently, based on the
comparison of the correlation of all the possible combinations of GDGTs with SST,
performed on a more extensive core-top data set (n = 396), a modified version of TEX86,
known as TEX 86L , was proposed [Kim et al., 2010]:
[GDGT  2]


TEX 86L  log

 [GDGT  1]  [GDGT  2]  [GDGT  3] 
L
The differences between these two indices are twofold. Firstly, the TEX 86
does not include
the crenarchaeol regioisomer, which could be difficult to quantify especially in subpolar
regions where its abundance is usually low, rendering possibly better applicability of the
L
modified equation in this realm. On the other hand, the logarithmic function of the TEX 86
indeed affords a better linearity to the regression line by amplifying the variability in the low
temperature range, but at the same time also magnifying its uncertainty.
The TEX86 values could be converted to SST by means of the empirical linear
regressions based on global sediment core-tops [Kim et al., 2008; Kim et al., 2010; Schouten
et al., 2002]. There have been some successful paleo SST reconstruction in the subtropical
25 Chapter 2 area such as in the Arabian Sea [Huguet et al., 2006b], the Mediterrenean Sea [Castañeda et
al., 2010; Huguet et al., 2011] and the Agulhas current system [Caley et al., 2011]. However,
the applicability of the global core-tops calibration of TEX86 and TEX 86L in the subpolar and
polar regions is more uncertain. Thus far there are few reported successful studies. In some
cases modified TEX86 indices and calibrations are proposed, such as demonstrated by Sluijs et
al. [2006] in the Arctic and Shevenell et al. [2011] at the Antarctic Peninsula. Furthermore,
the substantial scatter in the TEX86 – SST correlation [Kim et al., 2010] especially in the
lower temperature range of the calibration, suggests the possibility of additional factors that
might influence the TEX86 indices in these regions.
Therefore, this study was carried out to assess the applicability of the TEX86 and
TEX 86L indices in low-temperature environments. In light of the important role played by the
polar regions in the global climate, the emphasis of this study is focused on the mid and high
latitudes where the annual mean surface water temperatures are in the range of -2 to 17 °C. In
addition, this study also aims at providing more TEX86 core-top data from the Southern Ocean
and the Pacific, the two regions that still suffer from a lack of geographical coverage in the
present global core-top data set.
2.2. Materials and Methods
2.2.1. Sediment Extraction
The 160 marine sediment core-tops (0-1cm) analyzed in this study were obtained via
multicoring during several Alfred Wegener Institute (AWI) and German-Russian scientific
expeditions on various research vessels (R/Vs Polarstern, Sonne, Ivan Kireyev and Akademik
Boris Petrov) in the Southern Ocean, the North Pacific, the Fram Strait and the Arctic Ocean
(Figure 2.1).
All core-top sediments were freeze-fried and homogenized. The GDGT extraction of
the core-tops from the Southern Ocean (ANT-XXIII/9) and the Arctic (ARK-IX/4, ARKXI/1, Transdrift-1 and SIRRO 1997 - 2000) was done according to Müller et al. [1998].
Briefly, the sediments were subjected to 3 times of ultrasonication using UP200H sonic
disruptor probes in successively less polar solvent mixtures (dichloromethane/methanol).
From the North Pacific samples (SO202), the GDGTs were extracted using an DIONEX™
accelerated solvent extractor (DIONEX ASE 350), according to the NIOZ protocol [Schouten
26 Chapter 2 et al., 2002]: heating for 5 minutes at 100°C, static time of 5 minutes, 3 cycles and solvent
mixture of dichloromethane/methanol in the ratio of 9:1, v/v. After extraction, the total lipid
extracts were fractionated by column chromatography (SiO cartridges, Varian Bond-Elut)
using dichloromethane/methanol.
The core-tops from the central Arctic (ARK-XXII/2), the Fram Strait (ARK-XXI/1)
and the Southern Ocean core-tops other than those from ANT-XXIII/9, were subjected to
microwave assisted extraction as described by Fietz et al. [2011] by using a mixture of
dichloromethane:methanol (3/1, v/v). The extracts were then injected manually onto a Thermo
Surveyor HPLC system equipped with a Lichrosphere Silicon dioxide column (4.6 x 250 mm,
5 μm; Teknokroma) and a stainless steel inline filter (2 μm pore size). Compound class
fractionation was achieved running sequentially n-hexane, dichloromethane, and acetone.
Figure 2.1: Location of study sites in the global ocean. Black circles denote sites in this study, while previously
published data are marked by different symbols: black crosses - Kim et al. [2008] and Kim et al. [2010]; black
asterisks - Ho et al. [2011]; open triangles – Shevenell et al. [2011]; diagonal crosses – Leider et al. [2010]. Blue
and red lines illustrate the monthly sea ice extent (in March and September) incorporated in the ODV software
[Schlitzer, 2011] based on data from Walsh [1978] and Zwally et al. [1983]. Color bar indicates the altitude and
the depth of seafloor.
2.2.2. GDGTs Analysis
The GDGTs in samples from expeditions ANT-XXIII/9, SO202, ARK-XI/1, ARKIX/4, Transdrift-1 and SIRRO 1997 - 2001 were quantified using liquid chromatography with
a method derived from Hopmans et al. [2000]. The polar fractions were pre-filtered through a
4 m diameter PTFE filter (0.45 µm pore size) to prevent clogging in the column. Filtered
27 Chapter 2 samples were then dissolved in hexane:isopropanol (99:1; v/v) and were injected onto a high
performance liquid chromatography system (Agilent 1200 series HPLC system) coupled to an
Agilent 6120 MSD mass spectrometer, operating with atmospheric pressure chemical
ionization (APCI). The injection volume was 20 µL. A Prevail Cyano 3µm column (Grace,
150mm x 2.1mm) maintained at 30°C was used to separate the GDGTs. The injected samples
were eluted with a mixture of solvents, i.e., solvent A = hexane and solvent B = 5%
isopropanol in hexane. The mixture of solvents (solvents A:B in the ratio of 80:20 v/v) was
eluted isocratically for 5 min, then the volume of solvent B was increased linearly to 36% in
40 min. The column was back-flushed with 100% solvent B for 8 min after each analysis to
eliminate any compound that remains in the column. The spray chamber of the APCI-MS was
set in the following conditions: drying gas flow 5 l/min and temperature 350°C, nebulizer
pressure 60 psi, vaporizer gas temperature 450°C, capillary voltage -3 kV and corona current
+4 µA. The detection of the GDGTs was achieved by Selected Ion Monitoring (SIM) of the
[M+H]+ ions (dwell time 67 ms) in the m/z range of 1022 – 1302.
Meanwhile, the remaining core-tops (ANT-X/5, ANT-XI/2, ANT-XII/4, ANT-XX,
ANT-XXVI/2, ARK-XXI/1 and ARK-XXII/2) were analyzed as described by Fietz et al.
[2011]. The dry polar fractions were re-dissolved in hexane/n-propanol (99/1, v/v) and filtered
through 0.50 μm PTFE filters (Advantec). A Dionex P680 HPLC system coupled to a Thermo
Finnigan TSQ Quantum Discovery Max quadrupole mass spectrometer with an APCI
interface was used. The target compounds were separated with a Tracer Excel CN column
(0.4 x 20 cm, 3 μm; Teknokroma) equipped with a precolumn filter and a guard column. The
solvent program was modified from Schouten et al. [2007] and Escala et al. [2007]. Samples
were eluted with hexane/n-propanol at 0.6 mL / min. The amount of n-propanol was held at
1.5 % for 4 min, increased gradually to 5.0 % during 11 min, then increased to 10 % during 1
min and held at 10 % for 4 min, then decreased to 1.5 % during 1 min and held at 1.5 % for 9
min until the end of the run. The parameters of the APCI were set as follows to generate
positive ion spectra: corona discharge 3 μA, vaporizer temperature 400ºC, sheath gas pressure
49 mTorr, auxiliary gas (N2) pressure 5 mTorr and capillary temperature 200ºC. GDGTs were
monitored in SIM mode.
The samples preparation and GDGT analysis were carried out at the University of
Bremen, the Autonomous University of Barcelona and the Alfred Wegener Institute. We
contend that the results obtained from different laboratories using different analytical methods
should result in comparable data since it was reported that the TEX86 measurement is not
28 Chapter 2 significantly biased by extraction techniques or HPLC/APCI-MS set-ups [Schouten et al.,
2007].
2.2.3. TEX86 calculation and SST estimations
The values for TEX86 and TEX 86L are calculated following the equations as reported by
Schouten et al. [2002] and Kim et al. [2010], respectively. These values are converted to SST
estimates using the latest global core-top TEX86 / TEX 86L - SST calibrations proposed by Kim et
al. [2010]. Additionally, we calculated the Branched and Isoprenoid Tetraether Index (BIT)
according to Hopmans et al. [2004], which was proposed as a qualitative indicator for fluvial
terrigenous GDGTs input (cutoff value suggested to be 0.3 by Weijers et al. [2006]).
2.2.4. Environmental data
The objectively analyzed climatological data utilized in this study, i.e., water
temperature at different depths, are extracted from the World Ocean Atlas 2009 (WOA09)
[Locarnini et al., 2010]. The WOA09 dataset is preferred (over the NSIPP AVHRR
Pathfinder and Erosion Global 9 km SST Climatology dataset used in Kim et al. [2010] which
does not include subsurface water temperature data, see Figure 1.11) in this study on the
grounds that all the environmental data used to discuss our TEX86 values should be from a
consistent source.
For comparison of our index data with seasonal mean SST, we used seasons as defined
by the WOA09 (i.e., Northern/Southern hemisphere spring/autumn: AMJ, summer/winter:
JAS, autumn/spring: OND, winter/summer: JFM) [Locarnini et al., 2010]. The water
temperatures at various depths (e.g., base of thermocline) are determined manually from the
temperature-depth profile at each study site using the ODV software [Schlitzer, 2011]. Instead
of a fixed depth, we opt to compare TEX86 values with water temperatures at a particular
physical or chemical hydrographic boundary. This is because our study sites vary
considerably in terms of water depths (20m to 5500m), hence a fixed depth (e.g., 200m) could
correlate to very different hydrographic features, potentially introducing more scatter in the
correlation of the water temperature with the TEX86 values. In fact, if there is indeed a
significant subsurface GDGTs export, it is more likely to occur at a hydrographic boundary,
29 Chapter 2 physical or chemical, rather than a fixed depth, due to specific biological niche development
or physical factors such as preferential detention of lipids at the interface of water masses with
large density difference. Therefore, three water depths are chosen, i.e., the base of
thermocline, the base of oxycline, and the bottom waters close to the water-sediment interface
(defined for the purpose of our study as the deepest temperature datum available for a given
site in WOA09).
2.3. Results
2.3.1. Relationship between individual GDGTs
Figure 2.2: Principal component analysis on the individual GDGTs in the compilation of the data in this study
and the published core-top data by Kim et al. [2008], Kim et al. [2010] and Ho et al. [2011].
The PCA on the compilation of our data, the global core-top data set from Kim et al.
[2008; 2010] and the Pacific data set from Ho et al. [2011] spanning a large temperature range
(-2 to 30°C), shows that in general the relative relationship between individual GDGTs at our
core sites are not different from that of the global ocean. The exceptions are five North Pacific
samples (Cluster 1, Figure 2.2) and the Red Sea data set (previously discussed in Kim et al.
30 Chapter 2 [2008]) that form separate clusters. The first two principal components explain approximately
92% of the variance (PC1: 70 % and PC2: 22 %), similar to the values reported in the global
core-top study [Kim et al., 2010]. On the Component 1 axis, all the GDGTs included in the
TEX86 indices are positively correlated, while GDGT-1 and GDGT-2 are negatively
correlated with GDGT-3 and crenarchaeol regioisomer on the Component 2 axis. This pattern
of GDGTs distribution is essentially the same as reported in the latest global core-top study
[Kim et al., 2010].
2.3.2. Fractional abundance of GDGTs
In a broad sense, the fractional abundance of GDGTs increases with the WOA09
annual mean SST, except the GDGT-0, which exhibits an opposite trend (see Figure 2.3).
Some of the Arctic marginal sea data reveal a different relationship with SST compared to the
Southern Ocean and the North Pacific data set, especially evident in terms of the relative
fractional abundance of GDGT-3 (Laptev Sea and Kara Sea), crenarchaeol (Laptev Sea and
Kara Sea) and crenarchaeol regioisomer (Fram Strait and Central Arctic). Meanwhile, the
distribution of GDGTs from Cluster 1 displays a completely different behavior relative to the
rest of the data set. The fractional abundances of GDGT-0, -1 and -2 are remarkably high
while those of crenarchaeol and crenarchaeol regioisomer are substantially lower than in the
rest of the data set.
2.3.3. TEX86 and TEX86L values
For a broader geographical coverage in the appraisal of the applicability of TEX86 and
TEX 86L thermometries in the subpolar and polar regions, we include also previously published
data [Ho et al., 2011; Kim et al., 2008; Kim et al., 2010; Shevenell et al., 2011] from regions
that are comparable in terms of SST (<17°C) and latitudes (>37°N/S) with those presented in
this study.
31 Chapter 2 Figure 2.3: Relationship between the fractional abundance of individual GDGTs and the WOA09 annual mean
SSTs, i.e., (a) GDGT-0, (b) GDGT-1, (c) GDGT-2, (d) GDGT-3, (e) Crenarchaeol, (f) Cren’ (abbreviation for
crenarchaeol regioisomer). Oceanic regions are marked by different symbols in different colors as explained in
the legends.
32 Chapter 2 Figure 2.4: Correlation plots of TEX86 and TEX86L values with annual mean sea surface temperature derived
from World Ocean Atlas 2009 (WOA09). Oceanic provinces are indicated by different symbols in different
colors as explained in the legends. Filled symbols indicate data from this study, open symbols indicate published
data, grey-filled symbols indicate data with BIT >0.3. Abbreviations: SSI – summer sea ice; WSI – winter sea
ice, I – Indian, P – Pacific; A – Atlantic.
2.3.3.1. Correlation to temperature at various water depths
Overall, compared to its counterpart TEX86, the TEX 86L displays better correlation with
the SST (Figure 2.4). There is considerable scatter in the TEX86 data, especially those from
the Laptev Sea, the Fram Strait and the central Arctic, which form a separate cluster with
relatively high TEX86 values at low temperatures. On the other hand, the most pronounced
scatter in the TEX 86L data set is attributable mainly to the data from the Laptev Sea with BIT
values > 0.3 and two anomalous data from the Antarctic Peninsula. It is also notable that both
the TEX86 and TEX 86L values in the Barents Sea are invariant in spite of the large SST range
(0 – 7 °C) encompassed by these sites.
In addition to the sea surface (0 m), we also examine the correlation of TEX86 and
TEX 86L values with the temperature at 3 different water depths (Figure 2.5), namely the base
of the seasonal thermocline, the base of the oxycline (where the oxygen level is the lowest in
33 Chapter 2 the water column), and the bottom waters (deepest datum available for a given site in the
WOA data set). There is no strong linear relationship (low r2 values below 0.05) between both
the TEX86 and TEX 86L values with the temperature of bottom waters and the base of the
oxycline. On the contrary, a better correlation, comparable to that of the SST, is found for the
temperature at the base of the seasonal thermocline.
2.3.3.2. Correlation to SST of various seasons
There is no difference between the correlation of TEX86 values with the annual mean
SST (Figure 2.4) and that of the seasonal SSTs (Figure 2.6a – 2.6d). The same is true for the
TEX 86L data set (Figure 2.6e – 2.6h), with the exception of the correlation with the summer
SST, which is slightly poorer than those of the other seasons and the annual mean.
2.3.4. Residuals of SST estimates
The residual of SST estimate is defined here as the difference between the estimated
SST and the atlas SST. The residuals vary depending on the calibration used. Here we assess
how the SST estimates derived from the global core-top calibrations compare to the
climatological SSTs. The standard errors of estimates for the latest global core-top
calibrations for TEX86 and TEX 86L (Kim et al. [2010], n = 396) are 5.2°C and 4°C,
respectively.
The standard deviation of the TEX86 derived SST estimates (relative to annual mean
SST) is 6.1°C (Figure 2.7a). Although the residuals seem to become increasingly negative as
SST increases, a large majority of the SST estimates are within the standard error of
estimates. At the low temperature end, a separate cluster with deviant positive residuals
ranging between ~5 to 30°C is apparent, consisting mostly of data from the Arctic regions and
the Antarctic Peninsula. The global TEX86 calibration underestimates the SST at some
marginal seas such as the Sea of Japan, the North Sea, and the Irish Sea, resulting in negative
residuals.
34 Chapter 2 Figure 2.5: Correlation plots of TEX86 and TEX86L values with water temperatures derived from World Ocean Atlas 2009 (WOA09) at various depths. The water depths used in
the plots are as follows: Thermocline – the base of seasonal thermocline, Oxycline – the base of the oxycline , Bottom – the deepest water depth datum available for the site.
Oceanic provinces are indicated by different symbols in different colors as explained in the legends. Filled symbols indicate data from this study, open symbols indicate published
data, grey-filled symbols indicate data with BIT >0.3. Abbreviations: SSI – summer sea ice; WSI – winter sea ice, I – Indian, P – Pacific; A – Atlantic.
35 Chapter 2 Figure 2.6: Correlation plots of TEX86 and TEX86L values with mean seasonal sea surface temperature derived
from World Ocean Atlas 2009 (WOA09). Grey lines illustrate the latest global core-top calibrations [Kim et al.,
2010]. Oceanic provinces are indicated by different symbols in different colors as explained in the legends.
Filled symbols indicate data from this study, open symbols indicate published data, grey-filled symbols indicate
data with BIT >0.3. Abbreviations: SSI – summer sea ice; WSI – winter sea ice, I – Indian, P – Pacific; A –
Atlantic.
36 Chapter 2 Figure 2.7: The residuals of temperature estimates derived from the latest global core-top calibrations for TEX86 and TEX86L [Kim et al., 2010]. SST residuals are defined as the
subtraction of World Ocean Atlas 2009 (WOA09) annual mean/summer/winter SST from the estimated SST. The grey lines represent the standard error of estimates for the
calibrations. Oceanic provinces are indicated by different symbols in different colours as explained in the legends. Filled symbols indicate data from this study, open symbols
indicate published data, grey-filled symbols indicate data with BIT >0.3. Abbreviations: SSI – summer sea ice; WSI – winter sea ice, I – Indian, P – Pacific; A – Atlantic.
37 Chapter 2 On the other hand, the residuals of the TEX 86L inferred SST estimates (relative to the
annual mean SST) have a standard deviation of 6.5°C (Figure 2.7b). Some data from the
Arctic form a separate cluster with large positive residuals, albeit to a smaller extent
compared to the TEX86 temperature residuals. Two of the SST estimates at the Antarctic
Peninsula are unrealistically low, with underestimation down to -60°C. Most of the SST
estimates from the open ocean settings are overestimated beyond the standard error of
estimates, including those from the Southern Ocean, the North Pacific and the Bering Sea.
Meanwhile, underestimation mostly occurs in the marginal seas such as the Irish Sea, the
Black Sea and the North Sea.
To detect any potential seasonal bias in the SST estimates, we also assess the residuals
of the SST estimates relative to the summer and winter SST (Figure 7c to 7f). The standard
deviation of residuals relative to the winter SST are comparable to those of the annual mean,
and are smaller than those of the summer SST. In general, the residuals of seasonal SST
estimates display a similar data distribution as those of the annual mean SSTs, with
amplification in the extent of underestimation, especially in the temperature range of 15 –
23°C.
2.4. Discussion
2.4.1. GDGT contribution from methanogenic archaea
The PCA on individual GDGTs reveals that 5 samples from the North Pacific
(Subarctic Front and Alaska Gyre) and the Bering Sea have a distinctive GDGT signature
(Figure 2.2). On a closer inspection, these samples contain elevated abundance of GDGT-0,
and to a lesser extent also of the GDGT-1 and 2 (Figure 2.3), relative to crenarchaeol, the
hallmark GDGT for marine Thaumarchaeota [Sinninghe Damsté et al., 2002b]. The atypical
GDGT distribution in these samples suggests a contribution from an archaeal community
other than the pelagic Thaumarchaeota. Soil derived and fresh water related GDGTs are
unlikely to be the cause, since these sites are far from the coast, hence receive insignificant
riverine input of terrigenous GDGTs and are not in the vicinity of sea ice margin where fresh
water could be generated via ice melting. The abundance ratio of GDGT-0/crenarchaeol at
these sites ranges between 8 and 12, which are remarkably high compared to the values
observed in typical marine sediment (0.2 – 2, Schouten et al. [2002]). It has been proposed
that a ratio GDGT-0/crenarchaeol > 2 is indicative of significant methanogen archaeal input in
38 Chapter 2 lake sediments [Blaga et al., 2009], based on the assumption that the main contributors of
GDGT-0 and crenarchaeol are the methanogens and the Thaumarchaeota, respectively.
Therefore, the atypical GDGTs distribution observed at Cluster 1 sites in the North Pacific
and the Bering Sea might be attributable to major contribution from the methanogens.
Nevertheless, some of the SST estimates are within the standard error of estimates (three out
of five for TEX86 and one out of five for TEX 86L ). At those sites with anomalously large
residuals, there is an opposite trend in the TEX86 and TEX 86L derived SST estimates, i.e., the
TEX86 inferred estimates are warm-biased while those inferred from the TEX 86L index are coldbiased. The warm-biased TEX86 derived estimates are in agreement with the finding of a
study in Lake Challa where the methanogenic archaeal GDGTs biased TEX86 values were
anomalously high [Sinninghe Damsté et al., 2009]. Nonetheless, we exclude the Cluster 1 data
from further discussion to avoid undue bias.
2.4.2. Correlation of TEX86 / TEX86L indices to subsurface temperature
Since the GDGTs are found throughout the water column [e.g., Karner et al., 2001], it
is tempting to speculate that the lack of correlation with SST could be due to a substantial
input of GDGTs from deeper waters, interfering with the surface water GDGT signals.
However, we do not observe much improved correlations at deeper water depths, such as at
the base of oxycline and the bottom waters at the site (Figure 2.5). The correlations with the
seasonal thermocline (10-200 m), on the other hand, seem to be better than those with the
SST, especially evident for the TEX86 index (Figure 2.5a and Figure 2.5d). However, this
observation probably arises as an artifact of the reduced data set for the TEX86 – thermocline
correlation, which does not include most of the deviant data from the Arctic shelf (due to the
difficulty in identifying this hydrographic barrier in the poorly resolved WOA09 data set for
shelf regions). Indeed, if the same reduced data set would be considered for the correlation of
the GDGT indices with both the seasonal thermocline and SST, comparable or better r2 values
would be obtained for the SST (0.22 and 0.33 compared to 0.24 and 0.25 for the TEX86 and
TEX 86L , respectively). This finding attests to the fundamental principle of the TEX86
paleothermometry, i.e., the GDGTs that end up in the sediment are mainly those of the nearsurface dwelling archaea, in spite of the fact that the archaea thrive throughout the water
column. This assumption is based on several lines of evidence. For example, Sinninghe
Damsté et al. [2002a] and Wuchter et al. [2005] found that the sedimentary TEX86 values
39 Chapter 2 reflect the SST in spite of higher GDGT abundance in mid water depths owing to a more
efficient export of GDGTs from the surface waters presumably via an active food web
[Wakeham et al., 2003]. In addition, Huguet et al. [2006a] reported that the TEX86 derived
SST estimates in the gut of decapods, one group of zooplankton, are in good agreement with
the SSTs, suggesting that the zooplanktons graze upon the archaea or on the particles the
archaea are associated with in the surface waters. Furthermore, the archaeal cells are less than
1µm in size and are neutrally buoyant [Könneke et al., 2005; Margot et al., 2002] which
prevents them from sinking efficiently on their own [Wakeham et al., 2003].
2.4.3. Deviant SST estimates in the Arctic
Figure 2.8: Close-up of the residuals of (a) TEX86 and (b) TEX86L derived SST estimates relative to World
Ocean Atlas 2009 (WOA09) summer SST in the Arctic. The blue and red lines denote the maximum and
minimum monthly sea ice extent respectively, incorporated in the ODV software [Schlitzer, 2011] based on data
from Walsh [1978] and Zwally et al. [1983]. Color bar illustrates the residuals in blue-red color scheme: red
indicates overestimation while blue indicates underestimation.
The TEX86 derived SST estimates for the Arctic are largely overestimated, to the
extent of 28°C warmer than the summer SST at the sites (Figure 2.7 and Figure 2.8). Some
of the warm biased SST estimates near the coast, e.g., off Svalbard and off Siberia, are
probably the consequence of a significant input of terrigenous GDGTs via river runoff,
interfering with the marine signals. However, in the Arctic marginal seas especially the
Laptev Sea, there is a seaward increase in the magnitude of the SST overestimation, peaking
in the vicinity of the sea ice margin (Figure 2.8). This warm bias near the sea ice margin does
40 Chapter 2 not occur in the Barents Sea, even though there is indeed a seaward increase in the
temperature residuals. Interestingly, the extreme warm bias is also observed in the TEX 86L
derived SST estimates. In general, the warm TEX86/ TEX 86L SST estimates are associated with
high fractional abundance of GDGT-3 or crenarchaeol regioisomer (Figure 2.3). Coincidentally, a previous reconstruction work in the central Arctic during the PETM has also
found anomalously high GDGT-3 [Sluijs et al., 2006] that resulted in high TEX86 values. The
authors interpreted the deviant GDGT-3 as terrestrially originated, and discarded them from
the TEX86 calculation. Nevertheless, since the warm bias in our core-top TEX86 inferred SST
estimates is more pronounced near the sea ice margin than near the river mouth, we are more
inclined to attribute the warm bias to the occurrence of the sea ice. It is plausibly due to a
substantial input of GDGTs from sea-ice related archaeal communities with a different
temperature adaptation mechanism in the lipids, which might have skewed the TEX86 values
at the offshore Arctic sites. These archaea probably produce lipids with additional GDGT-3
(Figure 2.3d), resulting in anomalously high TEX86 values. Potential candidates for these
archaeal communities may be associated with the Marine Group 1 Crenarchaeota in the Arctic
[Alonso-Sáez et al., 2008; Bano et al., 2004], unspecified archaea in the sea ice [Junge et al.,
2004], or polar-specific Crenarchaeota dominant in the upper halocline of the Arctic waters
[Kalanetra et al., 2009]. Given that we do not observe such deviously warm SST estimates at
the sea ice edges in the Southern Ocean and the coast off the Antarctica (Figure 2.7), it is
possible that these sea ice archaea are limited to the Arctic. At this stage, however, it is
impossible to unambiguously identify the group of archaea that contributes to the spuriously
high TEX86 inferred SST estimates at our sites without further genomic work. Nevertheless,
our results suggest that extreme caution should be used when interpreting TEX86 data in areas
potentially influenced by sea ice.
2.4.4. Effect of sedimentation setting and/or seasonality
In addition to the spuriously warm Arctic SST estimates, there is also considerable
scatter in the subpolar TEX86 / TEX 86L data sets. A closer inspection of the scatter plots of
TEX86 / TEX 86L vs. SST (Figure 2.4) reveals two groups, congregating above and below the
global calibration of Kim et al. [2010] respectively. In general, the data in the group above are
mostly from the deep open ocean setting (depth > 2000 m), while the group below consists of
mostly data from the marginal seas and/or continental shelves. The same trend could also be
41 Chapter 2 observed in the TEX86 temperature residuals plots (Figure 2.7). Most of the warm-biased
estimates (e.g., summer residuals within the standard error of estimates) are consisted of data
from the open oceans, while the cold-biased estimates (e.g., winter residuals within the
standard error of estimates) occur at the marginal seas and the continental shelves such as the
Black Sea and the Gulf of St. Lawrence. The contrasting bias (warm bias vs. cold bias) in the
SST estimates at open ocean and marginal seas might be the reason why we do not observe
substantial difference in the correlation with seasonal SSTs (Figure 2.6), as the contradicting
estimates might have canceled out the correlation to warm or cold seasons.
A combination of factors, including archaeal ecology, sedimentation regime and
seasonality, might have contributed to the diverging TEX86 / TEX 86L - SST relationship in
different settings. The microbial communities are known to differ between coastal and open
ocean regions, with latitude, and in regions where there is upwelling of mesopelagic waters to
the surface [Giovannoni and Vergin, 2012]. For instance, the cold-biased TEX86 / TEX 86L derived SST estimates in marginal seas might arise as a consequence of higher archaeal cell /
lipid abundances in winter, as reported for the North Sea [Wuchter et al., 2005; Wuchter et al.,
2006] and the Antarctic coastal waters [Church et al., 2003; Murray et al., 1998]. On the
other hand, the GDGT flux to the seafloor in open ocean setting (e.g., the Southern Ocean, the
North Pacific) might be more tightly linked to the primary production export, which is heavily
biased towards the warm seasons [Honda et al., 2002; Honjo et al., 2000] due to more
availability of light, hence result in warm biased SST estimates. Similar findings, albeit on a
regional scale, were observed previously in a seaward transect study in the Mediterranean
[Leider et al., 2010], in which the authors found that the TEX86-derived SSTs were
increasingly overestimated seaward. The authors invoked seasonality in the production of
planktic archaea, the nutrient conditions and particle loading in surface waters to explain their
findings. These mechanisms might be in play as well in the marginal seas and the open oceans
in the subpolar and polar regions, causing the scatter in the low and mid temperature range in
the data set. Furthermore, some of the underestimation at the continental margins may also be
attributable to lateral transport, such as in the Argentine Basin, where previous studies
suggested that the cold-biased alkenone-derived SSTs observed are due to the advection of
allochthonous alkenones by vigorous surface currents [Benthien and Müller, 2000; Conte et
al., 2006; Mollenhauer et al., 2006; Rühlemann and Butzin, 2006]. Nevertheless, it is beyond
the scope of this study to reconcile the discrepancy in different sedimentation settings, hence
42 Chapter 2 the confirmation of our hypotheses awaits future work, e.g., water column studies across
different sedimentation settings.
2.4.5. Implications for global core-top calibrations
The compilation of our new data set, together with the global data set of Kim et al.
[2010] (Red Sea data not included) and some recently published regional core-top data sets
such as those from the Pacific [Ho et al., 2011], the Antarctic Peninsula [Shevenell et al.,
2011] and the Mediterranean Sea [Leider et al., 2010] results in an impressive data set of 630
data points (Figure 2.9). However, the linear regression through this data set yield mediocre
r2 values of 0.67 and 0.68 for TEX86 and TEX 86L , respectively, which are inferior compared to
those of the previously published global calibrations (0.77 and 0.86 for TEX86 and TEX 86L ).
The poorer correlation in this compilation might be attributable to the large scatter in the low
and mid temperature range. As discussed in Section 2.4.2, the SST estimates in the Arctic
regions are spuriously warm, hence the addition of these data might potentially jeopardize the
linear correlation of these index values with SST (Figure 2.9). Indeed, once all the data from
the Arctic region (Central Arctic, Laptev Sea, Kara Sea, Fram Strait, Barents Sea, off
Svalbard) are removed, the correlation of TEX86 with SST improves considerably, resulting in
r2 value of 0.8, which is slightly better than the r2 value of the Kim et al. [2010] TEX86
calibration (0.77). The standard error of estimate for this calibration is 4.8°C, slightly smaller
than that reported for Kim et al. [2010] TEX86 calibration (5.2°C). Interestingly, the scatter in
the mid temperature range (~7 – 17°C) seem to be larger than that at the lower end of the
temperature range (excluding the Arctic data), probably attributable to different sedimentation
setting (open ocean vs. marginal sea / continental shelf, see Section 2.4.4). On the other hand,
the removal of the Arctic data does not affect the correlation of TEX 86L with SST. Even if the
apparent outliers (see Figure 2.9b; two data from the Antarctic Peninsula with anomalously
low TEX 86L values) are removed, the r2 value of TEX 86L – SST correlation is only marginally
improved to 0.73, compared to 0.86 of the Kim et al. [2010] TEX 86L calibration. The
deterioration in the TEX 86L - SST correlation results in a larger standard error of estimates
(5.6°C compared to 4°C reported for Kim et al. [2010] calibration).
43 Chapter 2 Figure 2.9: Compilation of the (a) TEX86 and (b) TEX86L data from this study and the published data set of Kim
et al. [2010], Kim et al. [2008], Leider et al. [2010], Ho et al. [2011] and Shevenell et al. [2011], as a function of
World Ocean Atlas 2009 (WOA09) annual mean SST. Oceanic provinces are indicated by different symbols in
different colours as explained in the legends. Filled symbols indicate data from this study, open symbols indicate
published data, grey-filled symbols indicate data with BIT>0.3. Published calibrations are marked by dashed
lines while the calibrations derived from our compiled data set are marked by solid lines. Abbreviations: SSI –
summer sea ice; WSI – winter sea ice, I – Indian, P – Pacific; A – Atlantic; Eq. – Equatorial; WPWP – West
Pacific warm pool; AP – Antarctic Peninsula.
44 Chapter 2 2.4.6. Applicability in (sub)polar regions and conclusions
Since its first proposal [Schouten et al., 2002], the TEX86 paleothermometry has been
applied in several paleo reconstructions in the polar regions, such as in the Arctic during the
PETM [Sluijs et al., 2006] and in the Antarctic Peninsula during the Holocene [Shevenell et
al., 2011]. However, these studies employed modified versions of TEX86 calibrations due to
the unsuitability of the global core-top calibrations at these sites. To shed more light on the
validity of TEX86 thermometry in these regions, here we analyzed an extensive set of core-top
L
samples to investigate the correlation of the TEX86 and TEX 86
values to the SST and the
predictive power of the presently available calibrations.
With regards to the global calibrations reported by Kim et al. [2010], the TEX86
calibration appears to be a better choice for the SST reconstruction in subpolar and polar
regions. It yields better SST estimates (relative to WOA09 SST), especially in the Southern
Ocean and the North Pacific, compared to the recommended proxy for sites with SST < 15°C
[Kim et al., 2010], i.e., the TEX 86L calibration, which results in estimates that are warmer than
the WOA09 summer SST in these regions (Figure 2.7). Moreover, the application of the
TEX 86L calibration on a sediment core in the Antarctic Peninsula resulted in large temperature
variations during the Holocene [Shevenell et al., 2011 Supplementary information], with SSTs
below freezing point (~-20°C) which are probably unrealistic for the site. Taken together,
these findings argue against the applicability of the TEX 86L paleothermometry in these regions.
We observed some regional differences in the TEX86 / TEX 86L - SST relationships in
the subpolar and polar regions. For instance, regardless of the indices or calibrations used, the
SST estimates are cold- / warm-biased at some marginal seas (e.g., Irish Sea, North Sea) /
most of the Arctic sites especially in the vicinity of the sea ice margin. Caution should thus be
exercised in interpreting the TEX86 / TEX 86L - derived SST estimates in these regions. In spite
of these differences, the compilation of our data (excluding the Arctic data) with other
published data sets (n = 482) resulted in linear regressions with reasonably good fit (r2 values
> 0.7). The addition of data yields contrasting consequences to the global calibrations, i.e.,
improving the TEX86 (r2 = 0.8 and standard error = 4.8°C, compared to 0.77 and 5.2°C for
L
Kim et al. [2010]) while worsening the TEX 86
(r2 = 0.73 and standard error = 5.6°C,
compared to 0.86 and 4°C for Kim et al. [2010]). The contradictory findings in the TEX86 and
L
core-top data complicate the effort to disentangle the underlying causes contributing
TEX 86
45 Chapter 2 to the scatter in the data set in subpolar and polar regions. Nevertheless, better TEX86-SST
correlation, together with smaller TEX86-derived temperature residuals in the subpolar and
polar regions, lead us to conclude that TEX86 is a better SST proxy for application in these
regions.
2.5. Acknowledgements
We would like to express our gratitude to the chief scientists, scientists and crew
members on board of R/V Polarstern, R/V Sonne, R/V Akademik Boris Petrov, and R/V Ivan
Kireyev for samples retrieval during various expeditions. P. Masqué and P. Cámara are
thanked for providing samples from the central Arctic (PS70 samples). Thanks also go to
Carme Huguet for stimulating discussions that improved the data interpretation. A SCAR
Fellowship 2010/2011 is acknowledged for financially supporting the scientific stay of S.L.
Ho at the Universitat Autònoma de Barcelona. POLMAR Graduate School (Helmholtz
Association) is acknowledged for funding S.L. Ho’s PhD study at the Alfred Wegener
Institute for Polar and Marine Research (Bremerhaven).
46 Chapter 3 Chapter 3: Sea surface temperature variability in the Pacific sector of the
Southern Ocean over the past 700 kyr
Sze Ling Ho1*, Gesine Mollenhauer1, Frank Lamy1, Alfredo Martínez-Garcia2, Mahyar
Mohtadi3, Rainer Gersonde1, Dierk Hebbeln3, Samuel Nunez-Ricardo4,5, Antoni RosellMelé6,7, Ralf Tiedemann1
1
Alfred Wegener Institute for Polar and Marine Research, P.O.Box 12 01 61, 27515 Bremerhaven, Germany.
[*Corresponding author: [email protected]]
2
Geological Institute, Swiss Federal Institute of Technology Zürich, NO G 55, Sonneggstrasse 5, CH-8092
Zürich, Switzerland.
3
MARUM-Center for Marine Environmental Sciences, University of Bremen, Leobener Strasse, D-28359
Bremen, Germany.
4
Department of Zoology, Facultad de Ciencias Naturales y Oceanográficas, Universidad de Concepción, Casilla
160-C, 4070386 Concepción, Chile.
5
Presently at Programa de Biología, Facultad de Ciencias Básicas, Universidad del Magdalena, Cra. 32, No. 2208, P.O.Box 2-121630, Santa Marta D.T.C.H., Colombia.
6 Institució Catalana de Recerca i Estudis Avançats, Barcelona, Catalonia, Spain.
7
Also at Institut de Ciència i Tecnologia Ambientals, Universitat Autònoma de Barcelona, 08193 Bellatera,
Spain.
(In press in Paleoceanography)
3.0. Abstract
In spite of the important role played by the Southern Ocean in global climate, the few
existing paleoceanographic records in the east Pacific sector do not extend beyond one
glacial-interglacial cycle, hindering circumpolar comparison of past sea surface temperature
(SST) evolution in the Southern Ocean. Here we present 3 alkenone-based Pleistocene SST
records from the subantarctic and subtropical Pacific. We use a regional core-top calibration
data set to constrain the choice of calibrations for paleo SST estimation. Our core-top data
confirm that the alkenone-based UK37 and UK’37 values correlate linearly with the SST, in a
similar fashion as the most commonly used laboratory culture-based calibrations even at low
temperatures (down to ~1°C), rendering these calibrations appropriate for application in the
subantarctic Pacific. However, these alkenone indices yield diverging temporal trends in the
Pleistocene SST records. Based on the better agreement with 18O records and other SST
records in the subantarctic Southern Ocean, we propose that the UK37 is a better index for SST
reconstruction in this region than the more commonly used UK’37 index. The UK37-derived
SST records suggest glacial cooling of ~8°C and ~4°C in the subantarctic and subtropical
47 Chapter 3 Pacific, respectively. Such extent of subantarctic glacial cooling is comparable to that in other
sectors of the Southern Ocean, indicating a uniform circumpolar cooling during the
Pleistocene. Furthermore, our SST records also imply massive equatorward migrations of the
Antarctic Circumpolar Current (ACC) frontal systems and an enhanced transport of ACC
water to lower latitudes during glacials by the Peru-Chile Current.
3.1. Introduction
The Southern Ocean plays a key role in global climate via its influence in the
meridional overturning circulation [Marshall and Speer, 2012] and the global carbon cycle
[Fischer et al., 2010]. Knowledge of past changes in this ocean is therefore essential for a
better understanding of its mechanistic link to the global climate, and ultimately contributes to
improving the prediction of future climate change via modeling efforts. In this regard, sea
surface temperature (SST), as the interface between the ocean and the atmosphere, is an
indispensable boundary parameter in driving global climate models. Our present
understanding of Pleistocene SST evolution in the Southern Ocean is mostly derived from
sediment records in the Atlantic sector [Martínez-Garcia et al., 2009; Schneider-Mor et al.,
2008], the Indian sector [Howard and Prell, 1992] and the Southwest Pacific [Pahnke et al.,
2003; Schaefer et al., 2005]. The eastern Pacific sector of the Southern Ocean, on the other
hand, is a less studied region. The few existing high-resolution marine archives spanning one
glacial cycle off Chile at ODP Site 1233 (41°S) indicate a dramatic equatorward shift (7-10°)
of the Southern Ocean current systems [Verleye and Louwye, 2010] and substantial glacial
cooling of 5 to 7°C based on a coccolithophorid transfer function and alkenones [Kaiser et al.,
2005; Lamy et al., 2004; Saavedra-Pellitero et al., 2011]. Further south at 53°S, an alkenonebased SST record off the Strait of Magellan [Caniupán et al., 2011] displays glacial cooling
of up to ~8°C. Meanwhile, a time-slice study of the LGM at the East Pacific Rise using a
foraminiferal transfer function indicates a smaller amplitude of glacial-interglacial SST
changes of 2 to 5°C between 48°S and 57°S [Luz, 1977]. The few South Pacific data in a
circumpolar compilation from the subantarctic and the Antarctic zones of the Southern Ocean
based on siliceous microfossil records [Gersonde et al., 2005] suggest less severe glacial
cooling (~1.5°C) in the Pacific compared to the other sectors during the Last Glacial
Maximum (LGM). Notably, all these records do not extend beyond the last two glacialinterglacial cycles, hindering the comparison of temperature evolution in different sectors of
48 Chapter 3 the Southern Ocean and Antarctica on orbital time-scales. The lack of paleo SST records in
the subantarctic Pacific also precludes the examination of the SST gradients between low and
high latitudes, from which the latitudinal migration of the oceanic frontal systems and the
advection of the vigorous eastern boundary current, i.e., the Peru-Chile Current (PCC), could
be inferred. The transport of subantarctic cold water by the PCC to the tropics could influence
the SST in the cold tongue especially during glacial periods, as demonstrated by foraminiferal
census data and a simple heat model [Feldberg and Mix, 2002, 2003].
For the evaluation of the SST gradient, it would be ideal if the individual SST records
were derived from the same proxy and calibration in order to minimize the discrepancy that
might arise from dissimilar habitat depth and/or sensitivity of biological proxies to
environmental changes. In this work, we employ the most commonly applied organic
geochemical SST proxy, i.e., the alkenone paleothermometry. It is based on the relative
distribution of di-, tri- and tetra-unsaturated long-chain alkenones consisting of 37 carbon
atoms, generally known as C37:2, C37:3 and C37:4, respectively. The degree of alkenone
unsaturation is a function of growth temperature of the precursor, i.e., haptophyte algae. An
index known as UK37 (= [C37:2 – C37:4] / [C37:2 + C37:3 + C37:4]) has been proposed to quantify
the degree of unsaturation [Brassell et al., 1986], and it was later simplified to UK’37 (= [C37:2]
/ [C37:2 + C37:3]) since the C37:4 alkenones are often absent in open ocean sediments where
overlaying SSTs are higher than 12°C [Prahl and Wakeham, 1987]. Over the years, work has
been mainly focused on the simplified UK’37 index, which is applicable to most parts of the
global ocean. However, alkenone-derived glacial SSTs that are warmer than those of the
interglacial have been observed in the Sea of Okhotsk [Harada et al., 2006] and the Northeast
Atlantic [de Vernal et al., 2006; Rosell-Melé and Comes, 1999], raising doubts about the
applicability of alkenone paleothermometry at high latitudes. Another potential caveat, i.e.,
the non-linearity of the relationship of UK’37 index and SST at low temperatures (<6°C), has
also been suggested [Conte et al., 2006; Rosell-Melé, 1998; Rosell-Melé et al., 1994; Sikes
and Volkman, 1993]. It is still debatable whether UK37 or UK’37 is the more appropriate SST
proxy at high latitudes due to the lack of data in this region, especially in the Southern Ocean.
In this study we revisit the alkenone paleothermometry at the lower end of the
temperature range and assess the applicability of the alkenone indices by using regional
surface sediments. We present three SST records to investigate the temporal pattern and the
amplitude of the paleo SST evolution in the South Pacific along the latitudinal range of the
PCC spanning both subtropical and subantarctic oceanic zones. Based on our SST
49 Chapter 3 reconstruction, we infer the latitudinal migration of the oceanic fronts and discuss their
paleoclimatic implications.
3.2. Oceanographic setting
Figure 3.1: Location of sites and major oceanic currents discussed in this work. For the purpose of this study,
the subantarctic region is defined as the waters between the Subtropical Front and the Subantarctic Front. Blue
circles denote the sites of the core-top data in the regional alkenone unsaturation calibration, while blue open
diamonds denote the sites where alkenones were below detection limit. Black triangles denote the sites of SST
records used for discussion. In addition to the newly presented records (GeoB 3388-1, GeoB 3327-5 and
PS75/034-2), we also include several previously published SST records from the tropics (HY04 [Horikawa et
al., 2010] and the Southern Ocean (DSDP 594 [Schaefer et al., 2005], E47-018 [Howard and Prell, 1992],
PS2489-ODP Site 1090 [Martínez-Garcia et al., 2009] and ODP Site 1093 [Schneider-Mor et al., 2008]), in
addition to the Antarctic temperature record at EPICA Dome C [Jouzel et al., 2007]. Thin black lines indicate the
annual mean isotherms in degree Celsius (°C) derived from the World Ocean Atlas 2009 (WOA09). Orange
arrows indicate major surface currents, and colored lines illustrate the oceanic frontal system [after Orsi et al.,
1995]. Abbreviations: WSI = winter sea ice extent; APF = Antarctic Polar Front; SAF = Subantarctic Front; STF
= Subtropical Front; PCC = Peru-Chile Current; ACC = Antarctic Circumpolar Current; CHC = Cape Horn
Current.
The Peru-Chile Current (PCC; also known as Peru Current, Chile-Peru Current, and
Humboldt Current) and the Antarctic Circumpolar Current (ACC) are the main features of the
surface circulation in the Southeast Pacific (Figure 3.1). The eastward flowing ACC is driven
50 Chapter 3 by the intense mid-latitude Southern hemisphere westerly winds (Westerlies). Thus, its
latitudinal migration is closely related to the wind forcing [Orsi et al., 1995]. The circumpolar
transport of the ACC is approximately 107 Sv, with most of the transport occurring in the
Subantarctic Front (SAF) and the Antarctic Polar Front (APF) [Cunningham et al., 2003]. The
impingement of the northern part of the ACC onto the South American continent leads to a
bifurcation around 43°S, yielding a vigorous equatorward branch (PCC) and a weaker
poleward branch (Cape Horn Current, CHC) [Strub et al., 1998]. The PCC flows northward
along South America and is deflected away from the coast at around 5°S, feeding the cold
PCC water into the South Equatorial Current which flows westward as the equatorial cold
tongue between 10°S and 4°N [Wyrtki, 1965]. Meanwhile, the CHC moves along the coastal
region of southernmost Chile, mixing the subantarctic water with low salinity regional water
and transporting this modified ACC water to the Atlantic Ocean via the Drake Passage
[Chaigneau and Pizarro, 2005]. The Westerlies shift northward in the winter as a result of
seasonal fluctuations of sea ice around Antarctica [Kidston et al., 2011]. Modern day austral
winter is also marked by more vigorous advection of cold water towards the tropics and a
larger temperature gradient between low and high latitudes.
Our South Pacific core-top sites are located between the Subtropical Front (STF) and
the APF, where the modern day annual mean temperatures of the overlaying surface waters
are in the range of ~1 to ~12°C. Our long piston core sites are well suited for studying the
open ocean PCC, as they are beyond the direct influence of the intense coastal upwelling that
is confined within 50-60 km of the shoreline [Strub et al., 1998]. Sites GeoB 3327-5 and
PS75/034-2 are located at the northern extent of the ACC, in sensitive regions where the
latitudinal movement of the ACC is expected to be registered. The southernmost site
PS75/034-2 is located ~7° and ~9° north of the modern day mean location of the SAF and the
APF, respectively. Site GeoB 3388-1 lies within the flowpath of the PCC, thus the SST
changes here reflect the extent of the cold water advection by the PCC.
3.3. Materials and Methods
3.3.1. Materials
We analyzed 34 core-top samples (Figure 3.1) recovered by multicorer from the
Pacific sector of the Southern Ocean between the STF and the APF, but alkenones were
51 Chapter 3 detected at only 13 sites (see Table 3.1 for coordinates). Two piston cores in the subantarctic
Pacific sector of the Southern Ocean and one piston core from the subtropical South Pacific
were analyzed in this study (Figure 3.1). Core GeoB 3327-5 (43°14’S, 79°59’W, 3534 m
water depth, 900 cm length) and core GeoB 3388-1 (25°13’S, 75°31’W, 3558 m water depth,
710 cm length) were retrieved during the R/V Sonne cruise 102 [Hebbeln et al., 1995], while
core PS75/034-2 (54°22’S, 80°05’W, 4425 m water depth, 1808 cm length) was collected
during the Alfred Wegener Institute expedition ANT XXVI/2 with R/V Polarstern [Gersonde,
2011]. The sediments of core GeoB 3327-5 alternate between clayey foraminifera and clayey
foraminifera nannofossil ooze, while core GeoB 3388-1 consists of mainly nannofossil ooze.
Core PS75/034-2, on the other hand, due to its location below the carbonate compensation
depth, consists of mainly siliceous clay and is barren of foraminifera. The sampling intervals
for core GeoB 3327-5 and GeoB 3388-1 were 5 cm throughout the core. Core PS75/034-2
was sampled every 10 cm throughout the core and every 5 cm in the section between 200 cm
and 310 cm.
Table 3.1: Site information and alkenone index values of the Southern Ocean core-top samples retrieved via
multicoring.
Site
Longitude
Latitude
WOA09 Annual mean SST (°C)
PS75/034-1
80.09 °W
54.37 °S
6.71
K
U
37
0.213
K’
37
U
0.256
PS75/104-2
174.53 °E
44.77 °S
11.74
0.371
0.411
PS75/099-1
177.27 °E
48.26 °S
9.24
0.294
0.294
PS75/072-3
151.22 °W
57.56 °S
1.89
0.007
0.086
PS75/082-2
158.36 °W
59.04 °S
1.52
-0.012
0.081
PS75/098-6
179.01 °W
52.97 °S
7.47
0.217
0.275
PS75/105-1
174.62 °E
44.41 °S
11.74
0.486
0.486
PS75/095-6
174.43 °W
57.02 °S
4.95
0.178
0.214
PS75/101-2
175.88 °E
45.81 °S
11.12
0.296
0.324
PS75/080-2
157.64 °W
58.18 °S
2.08
-0.021
0.084
PS75/053-1
115.98 °W
60.77 °S
3.01
0.016
0.085
PS75/076-1
156.14 °W
55.53 °S
4.68
0.089
0.219
PS75/063-2
135.62 °W
58.90 °S
2.18
-0.007
0.093
3.3.2. 18O measurement on foraminifera
A Finnigan MAT 251 mass spectrometer coupled with a Kiel device inlet system was
used to measure the 18O composition of planktic Neogloboquadrina pachyderma (dextral
coiling) from the >150 µm size-fraction and benthic Cibicides spp. from the >212 µm sizefraction for core GeoB 3327-5. The measurements were performed on approximately 5-10
52 Chapter 3 individual tests. For all stable isotope measurements a working standard was used, which was
calibrated against VPDB (Vienna Pee Dee Belemnite) by using the NBS 19 standard.
Consequently, all isotopic data are relative to the PDB standard. Long-term analytical
standard deviation is ± 0.07 ‰ (Isotope Laboratory, Faculty of Geosciences, University of
Bremen).
3.3.3. Alkenone analysis
Sample preparation and alkenone analysis of cores GeoB 3327-5 and GeoB 3388-1
were carried out according to the procedure described by Müller et al [1998]. About 3 – 14 g
of freeze-dried and ground sediment samples were subjected to three times of sonication in
mixtures of methanol and dichloromethane with decreasing polarity. The supernatant was
then rinsed with deionized water and sodium sulfate, before being concentrated and passed
through a short bed of silica (Bond-Elut silica cartridge, Varian) to purify the fraction that
contained the alkenones. The fraction was then saponified to remove esters. The
quantification of alkenones was achieved using gas chromatography on an HP 5890, equipped
with a 60 m fused silica capillary column (DB-5 MS, Agilent) and a flame ionization detector.
The oven temperature was programmed to rise from 50 to 250°C at a rate of 25°C/min, then
to 290°C at a rate of 1°C/min, followed by 26 min of isothermal period, before being ramped
up to 310°C at a rate of 30°C/min and held constant for 10 min. Replicate analyses of
laboratory internal reference sediment suggest analytical errors of ~0.5°C for both alkenone
indices (UK37 and UK’37).
Piston core PS75/034-2 was analyzed at the Alfred Wegener Institute (Bremerhaven).
The extraction of organic compounds was accomplished using a Dionex ASE-200 pressurized
solvent extractor, with a mixture of methanol and dichloromethane in the ratio of 1:9. Similar
to the treatment for the GeoB cores, the total extract was separated into 3 fractions via silica
gel fractionation using hexane, DCM and methanol, respectively. The fraction (eluted with
DCM) containing the alkenones was then concentrated and analyzed by gas chromatography
on an HP 6890 fitted with a flame ionization detector and a 60 m DB-1 MS column (Agilent).
The initial temperature in the oven was set to 60°C. After the injection of samples, the
temperature in the oven was ramped up to 150°C at a rate of 20°C/min, followed by a reduced
rate of heating at 6°C/min until the final temperature of 320°C was achieved and held
constant for 40 min. The alkenone fraction of this sediment core was pure enough for
53 Chapter 3 quantification without saponification. We did not observe any systematic differences in the
alkenone index values between saponified and untreated extracts in the six samples we tested.
They agreed within ±0.015 units and ±0.012 units for UK37 and UK’37, respectively,
corresponding to ±0.47°C and ±0.37°C using the culture calibrations of Prahl et al. [1998].
Reproducibility of the instrument is estimated to be 0.17°C based on replicate analysis of
alaboratory E. huxleyi culture extract.
South Pacific core-top samples were subjected to microwave-assisted extraction,
followed by compound class fractionation using a Thermo Surveyor HPLC system equipped
with a Lichrosphere Silicon dioxide column, according to the methods described by Fietz et
al. [2011]. The fraction containing alkenones (eluted with DCM) was saponified to remove
co-eluting esters, prior to analysis by gas chromatography (same GC system used for the
alkenone analysis of piston core PS75/034-2 described above).
The concentrations of sediment extracts were adjusted such that the amounts of
alkenones injected for each measurement were above threshold values (>5-10 ng) to avoid
unjust bias due to low concentrations. The threshold values were previously suggested by
Rosell-Melé et al. [1995], Sonzogni et al. [1997] and Villanueva and Grimalt [1996].
3.3.4. Alkenone-based indices and calibrations
The identification of alkenones was achieved by comparing chromatographic retention
times of the samples with those of standards. The alkenone-based index (UK37 and UK’37)
values were calculated according to the previously proposed equations given in Section 3.1.
In order to compare these two alkenone indices in downcore reconstructions and to compare
SST records spanning the tropics and the subantarctic Pacific, we need UK37 and UK’37
calibrations that are based on a common data set and covering the largest possible temperature
range. For this purpose, we opted to convert the index values into sea surface temperature
using the widely used E. huxleyi culture-based calibrations proposed by Prahl et al. [1988],
i.e., UK37 = 0.04 T – 0.104 (r2 = 0.98) and UK’37 = 0.034 T + 0.039 (r2 = 0.99), the latter being
statistically identical to those based on global core-top compilations [Conte et al., 2006;
Müller et al., 1998].
54 Chapter 3 3.3.5. SST gradient calculation
We calculated the SST gradient along the latitudinal range of the PCC using alkenonederived SSTs. We preferred the UK37 index over the commonly used simplified version that
excludes the C37:4 alkenone, i.e., UK’37 (see justification in Section 3.5.1). Considering the
complexity of the hydrography in the eastern equatorial Pacific (EEP), we selected an openocean site HY04 (4°02’N 95°03’W, Horikawa et al. [2010]) that is beyond the influence of
the east Pacific cold tongue and the Peru coastal upwelling to examine the equator-to-pole
SST gradients. A recent core-top calibration study of Kienast et al. [2012] suggests that the
alkenone unsaturation in the open ocean EEP conforms to the established global core-top
calibrations. For the sake of consistency in the comparison, we recalculated the UK’37-derived
SST estimates at site HY04 using the laboratory culture-based calibration of Prahl et al.
[1988], assuming that C37:4 alkenones are absent here. This results in similar UK37 and UK’37
values. The assumption is justified by the fact that C37:4 alkenones are numerically significant
at growth temperatures below 15°C [Prahl et al., 1988], and that the difference between UK37
and UK’37 is only significant at ~10°C [Rosell-Melé, 1998]. Indeed, the recalculated SST
estimates based on the UK37 index are within ±0.5°C of the original UK’37-derived SST record
reported in the literature (Figure 3.6) with exactly the same temporal trends. The SST records
were re-sampled every 2 kyr for the calculation of the gradients between sites.
3.4. Results
3.4.1. Stratigraphy
In order to obtain a consistent stratigraphic framework for all records in the SST
gradients calculation, we tuned all available benthic 18O records to the global benthic 18O
stack LR04 [Lisiecki and Raymo, 2005] using the software package AnalySeries 2.0 [Paillard
et al., 1996]. For this purpose, we revised the published age model of GeoB 3388-1 [Mohtadi
et al., 2006] which was previously aligned to the orbitally-tuned ODP Site 677 [Shackleton et
al., 1990]. Overall, the differences between the revised and the original age models are
minimal, with one exception during the time interval between 400 kyr and 500 kyr, especially
at the termination of MIS 12. The linear sedimentation rates (LSR) at site GeoB 3388-1
fluctuate between 2.2 and 0.3 cm kyr-1, with an average of less than 1 cm kyr-1 over the past
700 kyr (Figure 3.2).
55 Chapter 3 The age model of core GeoB 3327-5 was similarly generated via graphical tuning of
the Cibicides spp. benthic 18O record to the LR04 global benthic stack. According to the age
model, the record extends back to 513 kyr and spans the past five glacial-interglacial cycles
(Figure 3.2). Average sedimentation rate is 2.6 cm kyr-1 and the values range between 0.7 cm
kyr-1 and 4.4 cm kyr-1 without any drastic fluctuation. The only exception is a brief interval
during MIS 7, where sedimentation rates reach about 10 cm kyr-1, which may suggest
redeposition. However, there is no lithological indication for, e.g., turbidites during this
interval. A lack of chronological tie points for MIS 9 and part of MIS 8 arises as a result of
poor carbonate preservation.
In core PS75/034-2 carbonate preservation is poor, thus a benthic foraminifera-based
18O record could not be obtained. The attempt to use radiolarian biofluctuation for
chronological control [Hays et al., 1976] has also failed due to low abundance of
Cycladophora davisiana (0 – 2.5% throughout the core) [Cortese, unpublished data]. There
are no well-dated marine records in the subantarctic Pacific that would provide a reference
chronology for graphical tuning of the downcore oscillations in the physical properties (e.g.,
lightness, major elements, magnetic susceptibility). In the absence of other alternatives, we
graphically tuned the PS75/034-2 UK37 record to the temperature evolution registered in the
EPICA ice core at Dome C, Antarctica [Jouzel et al., 2007], based on the updated chronology
EDC3 [Parrenin et al., 2007]. Justification for the preference of the UK37 index over the UK’37
index is outlined in Section 3.5.2. The EPICA T record was adjusted by a 15-point moving
average smoothing prior to the graphical alignment to accommodate the much lower temporal
resolution in core PS75/034-2. Our EDC3-derived age model is supported by the shipboard
biostratigraphy based on diatom zonation (Thalassiosira lentiginosa) [Zielinski and Gersonde,
2002], i.e., ~178 kyr and ~350 kyr in our EDC3-based chronology correspond to the
boundaries of MIS 6/7 and 9/10 as indicated by the biostratigraphy. The fairly uniform linear
sedimentation rate throughout the core (1.4 – 3.5 cm kyr-1) (Figure 3.2) and the resemblance
in the general patterns between core PS75/034-2 and other Southern Ocean records (Figure
3.5) provide additional confidence in the stratigraphic framework. We adopted the original
age model of core HY04 [Horikawa et al., 2010], which is based on visual alignment of the
benthic foraminiferal 18O to the orbitally-tuned ODP Site 677 [Shackleton et al., 1990] for
the upper 420 kyr, and the lower part of the record to the LR04 global stack. There is no
significant temporal offset between the upper 420 kyr of this 18O record (on current time
scale) and the LR04 benthic stack.
56 Chapter 3 Figure 3.2: Age models and linear sedimentation rates at core sites GeoB 3388-1, GeoB 3327-5 and PS75/034-2.
Shaded bars indicate glacial intervals, and the black numbers in the bars represent the marine isotope stages.
Black triangles illustrate the stratigraphic tie points while the black diamonds mark the shipboard
biostratigraphic points based on diatom zonation. (a) Stratigraphic framework for core GeoB 3388-1 was revised
by graphical tuning of the benthic foraminiferal 18O record (red curve) to the benthic 18O stack LR04 (grey
curve) [Lisiecki and Raymo, 2005]. The previously published age model [Mohtadi et al., 2006] is represented by
the blue curve. (b) Linear sedimentation rate at site GeoB 3388-1 derived using the stratigraphic tie points based
on the benthic 18O record. (c) Stratigraphic framework for core GeoB 3327-5 was established by graphical
tuning of the benthic foraminiferal (Cibicides spp.) 18O record (red curve) to the benthic 18O stack LR04 (gray
curve) [Lisiecki and Raymo, 2005]. (d) Linear sedimentation rate at site GeoB 3327-5 based on the tuned benthic
18O record. (e) Stratigraphic framework for core PS75/034-2 was established by graphical tuning of the UK37
record (red curve) to the smoothed (15-points running average) EPICA T record at Dome C, Antarctica [Jouzel
et al., 2007; Parrenin et al., 2007] (gray curve). (f) Linear sedimentation rate at site PS75/034-2 based on tuning
to the EPICA T record.
57 Chapter 3 3.4.2. South Pacific core-top alkenone calibrations
As shown in Figure 3.3, both UK37 and UK’37 indices correlate linearly to annual mean
WOA09 SST (with r2 values of 0.94 and 0.93, respectively) for the temperature range of 1.5
to 11.7°C. These regressions are identical within estimation error to the extrapolated Prahl et
al. [1988] calibrations below 8°C.
Figure 3.3: Correlations of alkenone indices (a) UK’37 and (b) UK37 with temperature. The sediment core-top data
were calibrated against the WOA09 annual mean SST, while the E. huxleyi culture data of Prahl et al. [1988]
plotted against the growth temperature as reported in the original publication. Blue line illustrates the linear
regressions proposed by Prahl et al. [1988], with extrapolation for temperatures below 8°C. Red line denotes our
South Pacific core-top calibration (through the black circles).
3.4.3. Downcore SST estimates and planktic 18O values
3.4.3.1. Core GeoB 3388-1
At subtropical site GeoB 3388-1, the UK’37-derived SSTs for the past 700 kyr range
between 15°C and 21°C (Figure 3.4a). The index suggests that SST during MIS 12 is slightly
colder (~2°C) than the average glacial SST, while MIS 13 is the coolest interglacial.
Meanwhile, the UK37-inferred SSTs at site GeoB 3388-1 are in the range of 16°C to 22°C. The
amplitudes of glacial/interglacial SST variations in both UK37- and UK’37-derived records are
~6°C.
3.4.3.2. Core GeoB 3327-5
The UK’37-SST estimates are between ~5°C and ~14°C over the past 513 kyr at site
58 Chapter 3 GeoB 3327-5 (Figure 3.4c). While there is not much difference in the warmth of interglacials,
the UK’37-inferred estimates suggest strong variability in the severity of glacials, with SSTs
from ~5°C during MIS 10 to ~10°C during MIS 6. On the other hand, the UK37-derived
glacial-interglacial SST oscillations at site GeoB 3327-5 range between ~8°C and ~16°C,
without any substantial long-term trend in glacial cooling and interglacial warming.
Alkenones in the top of a multicore at this site register UK37- and UK’37-inferred SST estimates
of 15.4°C and 13.9°C, respectively.
The 18O values of planktic dextral-coiling N. pachyderma range between 1.1 to 3.1‰
(Figure 3.4b). There is a data gap between MIS 8 and MIS 10 because of carbonate
dissolution. The 18O values during MIS 11 are more enriched than those in other
interglacials. Some abrupt shifts towards more depleted values are recorded during MIS 11
and 12.
3.4.3.3. Core PS75/034-2
The overall SST variability suggested by the UK’37 index at site PS75/034-2 is between
~1ºC and ~8ºC, resulting in a glacial-interglacial amplitude of up to ~7°C (Figure 3.4d). The
UK’37 index indicates that MIS 10 is the coldest glacial, while MIS 5 is the warmest
interglacial. During the interval between MIS 16 and MIS 12, the UK’37-inferred glacialinterglacial cycles are not pronounced due to substantially smaller amplitude of SST
oscillations (~2°C compared to ~7°C after MIS 12). The UK’37-derived SST estimates for
these glacial intervals (especially MIS 16) are as warm as the SST estimates for the
subsequent interglacial intervals. The UK’37 index suggests a pervasive long-term trend in the
glacial cooling, i.e., the glacial SSTs decrease from MIS 16 to MIS 10, and increase thereafter
to MIS 6, followed by a colder MIS 2. On the other hand, the UK37-derived SSTs at site
PS75/034-2 range between ~1°C and ~10°C over the past 700 kyr (Figure 3.4d). According
to the UK37-derived SST estimates, the severity of glacial SSTs does not vary substantially at
site PS75/034-2. MIS 10 is slightly warmer (~2°C) than the other glacial periods, while MIS 5
and MIS 13 stand out as the warmest and coolest interglacials, respectively. The SST
estimates inferred from the UK’37 and the UK37 indices for the top of a multicore at this site are
6.4°C and 7.9°C, respectively.
59 Chapter 3 Figure 3.4: Planktic 18O and alkenone-based SST records. Shaded bars indicate glacial intervals and the black
numbers in the bars represent the marine isotope stages. Gray bars denote modern day maximum and minimum
SSTs derived from WOA09. (a) SST records derived from alkenone based indices, i.e., UK37 (blue) and UK’37
(red) at site GeoB 3388-1. (b) Planktic 18O of dextral-coiling N. pachyderma at site GeoB 3327-5. Poor
carbonate preservation result in a data gap from MIS 8 to MIS 9. (c) SST records derived from alkenone based
indices, i.e., UK37 (blue) and UK’37 (red) at site GeoB 3327-5. Filled circles indicate core-top data at the same site.
(d) SST records derived from alkenone based indices, i.e., UK37 (blue) and UK’37 (red) at site PS75/034-2. Filled
circles indicate core-top data at the same site.
3.5. Discussion
3.5.1. Alkenone-based calibrations for application in the subantarctic Pacific
Here we use the E. huxleyi culture-based alkenone calibrations from Prahl et al.
[1988] for SST reconstruction. While the UK’37-SST relationship of this calibration has been
confirmed by global core-top calibrations [Conte et al., 2006; Müller et al., 1998] with
60 Chapter 3 extensive data sets encompassing diverse biogeographic provinces and a wide temperature
range, the UK37-SST correlation has not been calibrated globally. Thus, the UK37-SST
relationship outside the calibration range (T<8°C) is unknown except for the North Atlantic
and the Nordic Sea [Bendle and Rosell-Melé, 2004; Bendle et al., 2005; Rosell-Melé et al.,
1994; Rosell-Melé et al., 1995]. Considering the low modern SST at our southern site
PS75/034-2 (WOA09 annual mean SST of 6.7°C), the paleo SST here, especially during
glacials, are likely to be well below the calibrated temperature range of the culture calibration
(8-25°C). To better constrain our choice of calibrations, we examine the alkenone index
values in the South Pacific surface sediments and find that firstly, the linearity of both UK37and UK’37-SST correlations holds even at low temperatures in the South Pacific, indicating
that both indices faithfully record modern SSTs in this temperature range. Secondly, the
sedimentary alkenone unsaturation-SST relationships in the South Pacific are comparable to
those observed in the E. huxleyi culture of Prahl et al. [1988], rendering these culture
calibrations suitable for application in this region. Indeed, the UK37 and the UK’37 calibrations
resulted in core-top SST estimates (8°C and 6°C at PS75/034-2; 15°C and 14°C at GeoB
3327-5) that are within the range of modern seasonal SSTs (see grey bars in Figure 3.4; 5 –
9°C at PS75/034-2 and 10 – 15°C at GeoB 3327-5). We refrain from using our own core-top
calibrations for downcore reconstruction because of their limited calibration range (~1-12°C)
which makes them inappropriate for the application in the subtropics for calculating the
meridional SST gradients.
We note that our finding is in contrast to that of the Southern Ocean core-top
calibration study of Sikes et al. [1997]. The better correlation in the UK’37-SST relationship (r2
value of 0.92 compared to r2 value of 0.76 for UK37-SST) led the authors to suggest that the
UK’37 is the better index for paleo SST reconstruction in the Southern Ocean. Application of
their calibrations at our sites yields core-top SST estimates (UK37 and UK’37: 14°C and 9°C at
PS75/034-2; 21°C and 16°C at GeoB 3327-5) that are warmer than those inferred from the
Prahl et al. [1988] calibrations. The warm bias is especially pronounced in the UK37-derived
estimates, which are substantially warmer than the modern day warmest month SST in
WOA09 (see grey bars in Figure 3.4; ~9°C at PS75/034-2; ~15°C at GeoB 3327-5). These
anomalously warm estimates produced by the core-top calibrations of Sikes et al. [1997], in
addition to a good match between our South Pacific core-top calibrations and the Prahl et al.
[1988] culture calibrations, led us to choose the latter calibrations to estimate paleo SSTs at
our study sites in the subantarctic Pacific.
61 Chapter 3 3.5.2. Assessing contrasting temporal trends in UK37- and UK’37-derived SST records
In alkenone-based SST records, the temporal trend is governed by the definition of the
index, while the amplitude of downcore variation and the absolute value are determined by
the calibration employed. In the subtropics (GeoB 3388-1), the SST patterns inferred from
both UK37 and UK’37 indices are similar, and their values are in agreement within 1.5°C. As
discussed in section 3.5.1, the strong linear relationship between both the UK37 and the UK’37
indices in the subantarctic surface sediments with the overlaying SSTs (i.e., comparable r2
values) imply that both indices may be used to obtain paleo SST estimates in the region
(Figure 3.3). However, downcore reconstructions yield a different picture, i.e., the indices
result in contrasting subantarctic SST patterns for cores GeoB 3327-5 and PS75/034-2
(Figure 3.4). For the past two glacial-interglacial cycles, the UK’37–derived SSTs display a socalled Type 1 [Schneider et al., 1999] alkenone SST record which is typical for the tropics
and the monsoon-influenced region, characterized by a relatively warm MIS 6 and the
occurrence of the coldest glacial SST in the middle or the inception of glacials. There is also a
warming trend of glacials from MIS 10 to MIS 6 in these subantarctic UK’37 records. On the
other hand, the UK37-derived SST records suggest little fluctuation in the severity of glacial
intervals and the MIS 6 is as cold as other glacial intervals (a Type 3 alkenone SST record
according to the definition of Schneider et al. [1999]), which shows more resemblance to the
global ice volume oscillations documented in the benthic 18O record. The differences in
temporal trends are especially clear for the time interval MIS 16 – 12 at our southernmost site
PS75/034-2, during which the UK’37-derived SSTs exhibit a reduced amplitude of glacialinterglacial SST variations due to relatively warm glacials, especially MIS 16 which is as
warm as interglacial MIS 11. However, the UK37 index record suggests that the glacial SSTs
during this time interval are consistent with those from other glacial intervals. Interestingly,
such observations are not limited to the South Pacific. As shown in Figure 3.5c, dissimilar
amplitudes of glacial-interglacial SST oscillations during MIS 12 – 16 are also evident in the
alkenone-derived SST records at PS2489-2/ODP Site 1090 in the mid-latitudes of the South
Atlantic [Martínez-Garcia et al., 2009; Martínez-Garcia et al., 2010], suggesting that this
divergene can be found throughout the Southern Ocean south of the Subtropical Front. To
determine which pattern is more realistic, we further compare our alkenone records with the
planktic 18O record at the same site, and with other subantarctic SST records from other
sectors of the Southern Ocean (Figure 3.5). Since the most outstanding divergence in the two
62 Chapter 3 different alkenone SST patterns is in the long-term trend of the glacial severity (interglacial
warmth is consistent), we focus our discussion on the cold intervals.
Contrary to the UK’37–based SST records, the planktic 18O records in the South
Pacific (GeoB 3327-5) and the South Atlantic (PS2489-2 / ODP Site 1090, see Venz and
Hodell [2002]) suggest minor oscillations in glacial severity. Apart from global ice volume
and SST, the planktic 18O records are also influenced by changes in sea surface salinity
(SSS). However, given the lack of any major freshwater sources in the vicinity of sites GeoB
3327-5 and PS2489-2 / ODP Site 1090, large perturbations to the SSS at these sites over the
past 700 kyr are unlikely. SSS here might be driven by an enhanced influence of low SST and
low SSS polar water mass during glacials. However, in such a scenario, the SSS variations
would be accompanied by concurrent changes in SST. Therefore we believe that SSS
variations are not the reason for the diverging trends between the planktic 18O and the UK’37
records.
In addition to a warming trend in glacial severity from MIS 10 to MIS 6, the UK’37
SST estimates for MIS 12, 14 and 16 are relatively warm at sites PS75/034-2 and PS2489 /
ODP Site 1090, even though MIS 12 and MIS 16 are known to be among the most severe
glacial stages during the Pleistocene [Lang and Wolff, 2011; Shackleton, 1987]. We note that
varying Pleistocene glacial severity is not physically impossible. Indeed, a SST record in the
subtropical Agulhas region suggested its occurrence [Bard and Rickaby, 2009]. Here, MIS 10
and 12 are substantially colder than other glacials in the past 800 kyr; but the glacialinterglacial cycles before MIS 12 are well-defined – unlike in the subantarctic UK’37 records.
Furthermore, the Agulhas core site is located north of the Subtropical Front, under the
influence of a completely different hydrographic setting (e.g., warm Agulhas current and
associated eddies) from that of the subantarctic Southern Ocean. These differences suggest
that the varying glacial severity trends in Bard and Rickaby [2009]’s Agulhas SST record and
the subantarctic UK’37 SST records are unrelated.
On the other hand, the glacial severity trends in UK37-derived SST records are in
agreement with the planktic 18O records at site GeoB 3327-5 and PS2489-2 / ODP Site 1090.
At the latter site, a summer SST record inferred from foraminiferal assemblages further
supports this pattern [Becquey and Gersonde, 2002, 2003] (Figure 3.5d). Similar patterns in
glacial severity over the past 700 kyr has been observed elsewhere in the subantarctic
Southern Ocean and Antarctica, such as ODP Site 1093 and ODP Site 1094 in the South
63 Chapter 3 Atlantic [Schneider-Mor et al., 2008], DSDP Site 594 off New Zealand [Schaefer et al.,
2005], South Indian [Howard and Prell, 1992] and Antarctic atmospheric temperature records
at EPICA Dome C [Jouzel et al., 2007] and Dome Vostok [Petit et al., 1999] (Figure 3.5e –
3.5h). These temperature records suggest that unvarying glacial severity is a pervasive
Pleistocene climatic feature in the Southern Ocean.
The better agreement of the temporal trend of the UK37 than the UK’37 SST records with
other surface proxy records in the same oceanic region suggests that the UK37-derived SSTs
are plausibly more realistic than the UK’37 estimates at these sites, even though the core-top
values of both indices correlate equally well with modern SSTs. Our findings agree with a
multi-proxy comparison study off the Iberian margin [Bard, 2001]. The author found that the
UK37-derived glacial coolings were more comparable with those derived from other proxies,
even though the core-top SST estimates inferred from both UK37 and UK’37 indices were
comparable with the observed annual average SST. These findings demonstrate that different
alkenone indices could result in diverging paleo SST patterns during the cold intervals even if
the core-top SST estimates suggested by both indices agree with the modern day SST. The
discrepancy in paleo SST patterns stems from the higher relative abundance of the C37:4
alkenones during the cold intervals. Having established that the UK37 index is a more suitable
SST proxy in the subantarctic Pacific (south of the Subtropical Front at ~30°S), we base our
stratigraphic framework of PS75/034-2 and the following discussion on the SST variations
and the meridional gradients on the UK37- derived SST records.
3.5.3. Southern Ocean SST evolution: circum-Antarctic comparison
High-resolution alkenone SST records off Chile (e.g., ODP Site 1233 and MD073128) suggest that the SST in the mid-latitude Southeast Pacific evolved in synchrony with
the atmospheric temperature at Antarctica on millennial time-scales over the past 70 kyr
[Caniupán et al., 2011; Kaiser et al., 2005; Lamy et al., 2004]. Due to the coarser temporal
resolution in our Pleistocene SST records, it is impossible to assess these millennial-scale
patterns. Instead, our SST records, especially the southern site, share first-order patterns on
glacial-interglacial timescale with the EPICA Dome C temperature record of Jouzel et al.
[2007]. There are, however, some minor differences compared to the Antarctic temperature
record, such as the absence of a luke-warm interglacial MIS 15 at site PS75/034-2, and a
cooling during MIS 3 at site GeoB 3327-5. Besides, unlike in the Antarctic temperature
64 Chapter 3 record, the Mid-Brunhes Event (~430 kyr) shift is not well expressed in our SST records from
the Southeast Pacific (Figure 3.5). This suggests an overprint of regional climate in our
subantarctic SST records on the background of glacial-interglacial climatic changes closely
linked to Antarctica. Meanwhile, other features such as the coolest MIS 13 and the warmest
MIS 5 in the past 700 kyr, and the smallest amplitude of termination during the MIS 14 – MIS
13 transition observed in our records are common in many marine and terrestrial records
[Lang and Wolff, 2011]. With the exception of a warmer-than-today MIS 5 and a colder-thantoday MIS13, the maximum SST estimates for other interglacials at sites GeoB 3327-5 and
PS75/034-2 are similar to modern day summer SST (Figure 3.4).
The intensity of Pleistocene glacial cooling (~8°C) at our subantarctic Pacific sites is
within the range of other subantarctic SST records derived from various proxies (Figure 3.5),
i.e., ~5°C in the South Indian [Howard and Prell, 1992], ~7 to 10°C in the Southwest Pacific
[Pahnke et al., 2003; Schaefer et al., 2005], and ~7 to 11°C in the South Atlantic [Becquey
and Gersonde, 2003; Martínez-Garcia et al., 2009], indicating that the Pleistocene glacial
cooling in the Southeast Pacific is comparable, if not stronger, than in other sectors of the
Southern Ocean. This is in contrast to the findings of Gersonde et al. [2005] in a circumAntarctic LGM SST study using siliceous microfossil transfer functions. The authors reported
a non-uniform glacial cooling in the Southern Ocean, with less cooling (~1°C) in the Pacific
compared to the Atlantic and Indian sectors (4-5°C). The discrepancy between this study and
our compilation may be due to the more climatically sensitive sites of the long Pleistocene
records (i.e., DSDP 594, GeoB 3327-5, PS75/034-2, MD97-2021). Alternatively, it could also
be
due
to
the
different
sensitivity
of
proxies
(siliceous
microfossils
vs.
geochemical/carbonaceous microfossils) or the fact that the South Pacific is under-represented
in their calibration database. Indeed, foraminiferal assemblage-based LGM time slice studies
suggest cooling of ~5°C in the subantarctic Southeast Pacific (111-123°W) [Luz, 1977] and
up to ~8°C in the Southwest Pacific [Barrows and Juggins, 2005], in better agreement with
our alkenone-based estimates than those derived from the siliceous microfossil transfer
functions.
If true, the substantial Pleistocene glacial cooling in the subantarctic Southeast Pacific
suggested by the alkenone paleothermometry is plausibly due to an extensive equatorward
migration of the Westerlies and the Southern Ocean frontal systems embedded within the
65 Chapter 3 Figure 3.5: Comparison of temperature records from the Southern Ocean and Antarctica based on different
proxies. Shaded bars indicate glacial intervals and the black numbers in the bars represent the marine isotope
stages. (a) Alkenone UK37-derived sea surface temperature record at site GeoB 3327-5 in the Southeast Pacific.
(b) Alkenone UK37-derived at site PS75/034-2 in the Southeast Pacific. (c) Alkenone-derived sea surface
temperature records based on the UK’37 index (light purple curve) and the UK37 index (dark purple curve)
[Martínez-Garcia et al., 2010; Martínez-Garcia et al., 2009] at site PS2489-2 / ODP Site 1090 in the South
Atlantic. (d) Foraminiferal transfer function-derived summer SST record [Becquey and Gersonde, 2002; 2003] at
site PS2489/ ODP Site 1090. The authors regarded the estimates for MIS 11 as an overestimation due to
preferential dissolution of cold-water species. (e) Diatom transfer function-derived summer sea surface
temperature record [Schneider-Mor et al., 2008] at ODP Site 1093 in the South Atlantic. (f) Foraminiferal
transfer function-derived winter SST record [Schaefer et al., 2005] at site DSDP 594 in the Southwest Pacific.
(g) Foraminiferal transfer function-derived winter SST record [Howard and Prell, 1992] at site E49-018 in the
South Indian. (h) Atmospheric temperature record registered in the EPICA ice core at Dome C, Antarctica
[Jouzel et al., 2007].
66 Chapter 3 ACC, superimposed on the generally colder climate during glacials. Such equatorward shift of
the oceanic systems might have occurred as a consequence of a massive northward sea ice
expansion by 5° to 10°, as suggested previously by various faunal-based sea-ice and IRD
records in the Southern Ocean [Becquey and Gersonde, 2002, 2003; Crosta et al., 2004;
Gersonde et al., 2005]. By using the present as an analogue for the past and assuming that the
SST ranges associated with the oceanic fronts during glacial intervals would remain the same
as modern day (~5°C in the SAF and ~2°C in the APF as in Figure 3.1), the average glacial
SST estimates for sites PS75/034-2 (~1°C) and GeoB 3327-5 (~9°C) imply that both the SAF
and the APF were located between 43°S and 54°S in the Southeast Pacific during glacials.
This suggests that these oceanic fronts underwent substantial equatorward migration of ~7°
(SAF) and ~9° (APF) during glacials and resided northward of site PS75/034-2. Such frontal
migrations are conceivable, considering that no shallow bathymetric feature stands between
site PS75/034-2 and the modern average latitudes of these oceanic fronts. Thus, no
topographic obstacle restricts the equatorward movement. In fact, frontal shifts (SAF and
APF) of such magnitude during the Pleistocene have previously been proposed for the
subantarctic Atlantic [Becquey and Gersonde, 2003] and the Southwest Pacific [Schaefer et
al., 2005; Wells and Okada, 1997].
Such massive equatorward shifts of the ACC and its associated fronts in the Southeast
Pacific may have important implications for the water transport through the Drake Passage. If,
for instance, the SAF and the APF, which transport the bulk of the water in the ACC system,
would be deflected equatorward within the PCC instead of flowing through the Drake Passage
as they do today, the transport to the South Atlantic would have been markedly reduced
during glacials. In fact, such a scenario was invoked by Gersonde et al. [2003] to explain the
intense cooling east of the Argentine basin during the LGM. The authors further hypothesized
that such changes in the transport through the Drake Passage, which is one of the “Cold Water
Routes” of the global thermohaline circulation, would have major implications for the global
climate development. Our records corroborate their hypothesis and further suggest that the
same mechanism might have occurred during all glacials prior to the LGM over the past 700
kyr.
67 Chapter 3 3.5.4. Meridional SST gradients: equatorward cold water transport
Considering the large latitudinal range covered by the study sites, the alkenoneinferred SST records might be affected by different biogeographic patterns or seasonality. For
instance, if the abundances of the alkenones or the source organisms (e.g., E. huxleyi) are
skewed towards the warm / cold season at high / low latitudes [Schneider et al., 2010], the
resulting SST gradient would be artificially reduced. Thus, our estimation of meridional SST
gradients is conservative and might be underestimated.
Figure 3.6: Meridional gradients of alkenone-inferred SSTs and mean annual insolation along the Southeast
Pacific and SST evolution in the tropical Pacific. Shaded bars indicate glacial intervals, and the black numbers in
the bars represent the marine isotope stages. The meridional SST gradients (a) between the tropics (HY04) and
the subantarctic (PS75/034-2) (b) between the subtropics (GeoB 3388-1) and the subantartic (PS75/034-2) (c)
between the tropics (HY04) and the subtropics (GeoB 3388-1) are derived from UK37 SST estimates calculated
using the E. huxleyi culture calibration of Prahl et al. [1988]. (d) Meridional mean annual insolation gradient
between 4°N and 54°S [Laskar, 1990] (e) Alkenone-based SST records at site HY04 [Horikawa et al., 2010].
For the SST gradient reconstruction, we recalculated the published SST using the UK37 calibration of Prahl et al.
[1988] (black curve) so that it is consistent with other SST records. The gray curve depicts the originally
published SST record by Horikawa et al. [2010]. 68 Chapter 3 Our results show that in contrast to the pronounced glacial cooling in the subantarctic
Pacific (~8°C), the amplitudes of glacial cooling decrease to ~4°C and ~1.5°C in the
subtropics (GeoB 3388-1) and the tropics (HY04) (Figure 3.6), respectively. The glacial SST
estimates in the subtropics (GeoB 3388-1) are 1-2°C colder than the modern SST associated
with the STF in the Southeast Pacific (~19°C), suggesting that the STF might have also
shifted equatorward along with the SAF and the APF, albeit to a smaller extent, and resided
slightly northward of our study site. The SST gradients between low and high latitudes (4°N
at HY04 and 54°S at PS75/034-2) are steeper during glacials than interglacials, and the
overall pattern resembles a mirror image of the high latitude SST record (see Figure 3.6). The
pattern holds even if other EEP SST records such as the ODP 846 (cold-tongue) and ODP
1239 (coastal upwelling) are used for gradient calculation. The more substantial glacial
cooling at the higher latitudes leads to steeper SST gradients between the subantarctic and the
subtropics, than those between the subtropics and the tropics. Notably, the smaller tropicalsubantarctic SST gradient during MIS 4 is of the same magnitude as those of MIS 8, 10, 12,
14 and 16, while the SST gradients are larger during MIS 2 and MIS 6. The finding of steeper
SST gradients between the tropics and mid-latitudes during glacials is consistent with the
observation of Kaiser et al. [2005] over the past 70 kyr in the Southeast Pacific. However,
their reconstruction suggested a slightly larger gradient (~1°C) during MIS 4 than during
LGM, in contrast to ours. The discrepancy stems from the less intense cooling during MIS 4
at site GeoB 3388-1 relative to other glacials. Alternatively, it might also be due to a
combination of other factors, including the lower temporal resolution in our records, records
derived from different proxies (foraminiferal census count and Mg/Ca ratio) used in the
gradient calculation of Kaiser et al. [2005], or different SST calibrations employed (UK’37 vs.
UK37). Notwithstanding, our records indicate that steeper meridional SST gradients during
glacials are a recurring feature in the Southeast Pacific over the past 700 kyr.
Several factors may contribute to the steeper high-to-low latitude gradients, including
the insolation gradient and local hydrographic dynamics. The temporal resolution of our SST
records is insufficient for determining the contribution of the local insolation gradient in
shaping the meridional SST gradient, based on the wiggle-matching of the SST gradients to
the insolation gradients (Figure 3.6). Besides, the subtropical site GeoB3388-1 might also be
influenced by filaments advected from the coastal upwelling off Chile, if the upwelling was
stronger in the past. This notion, however, cannot be rigorously tested by our SST records and
awaits future work based on more conservative water mass tracers. Alternatively, the steeper
69 Chapter 3 high-to-low latitude gradients during glacials might be linked to the vigor of the PCC. As
readily observable in the modern day SST contour map (Figure 3.1), site GeoB 3388-1 is
characterized by the advection of cold water from the south. It is conceivable that the steeper
gradients between this site and the tropics (site HY 04 is beyond the influence of the east
Pacific cold tongue) during the glacial periods are a result of enhanced cold water transport
via an intensified PCC. Increased influence of ACC-sourced water in the subtropical
Southeast Pacific has been inferred from enhanced glacial paleoproductivity, assuming that
the main nutrient source was supplied from the south via the PCC [Mohtadi and Hebbeln,
2004; Romero et al., 2006]. Increased transport by the PCC during glacials was invoked to
explain the higher abundance of ACC cold-water coccolithophorid and dinoflagellate species
at the mid-latitudes Southeast Pacific [Saavedra-Pellitero et al., 2011; Verleye and Louwye,
2010] and the increased cold-water foraminiferal abundance in the equatorial Pacific
[Feldberg and Mix, 2002; 2003]. In addition, it has also been proposed on the basis of a
steeper glacial meridional SST gradient at the equator, which suggested a northward shift of
the Equator Front - Intertropical Convergence Zone (ITCZ) during glacial periods [RincónMartínez et al., 2010]. Stronger cooling and intensification in the PCC transport (an eastern
boundary current) during the glacial periods might have resulted from enhanced Ekman
pumping from the subantarctic zone, as a response to an increase in wind strength and/or
northward migration of the Westerlies. Such changes in the southern Westerlies have been
inferred from some marine records [e.g., Mohtadi and Hebbeln, 2004; Stuut and Lamy, 2004].
Indeed, based on the conservation of energy, a stronger zonal circulation north of the
subantarctic zone could be deduced from steeper meridional gradients and an equatorward
contraction of the subtropical realm. Moreover, as mentioned in Section 3.5.3, an
equatorward deflection of the major ACC fronts (the SAF and the APF) would also contribute
to increased cold water transport via the PCC.
3.6. Conclusions
The empirical relationship of UK37- and UK’37 with SST in our South Pacific regional
core-top data set is similar to the commonly used calibrations derived from the laboratory E.
huxleyi culture of Prahl et al. [1988]. These linear relationships hold even at low temperatures
(down to ~1°C), suggesting that the temperature dependence of the alkenone indices is not
lost at low temperatures in the Southern Ocean. This finding indicates that both alkenone
70 Chapter 3 indices are suitable for reconstructing SST at our cold subantarctic sites. However, these
indices result in dissimilar SST patterns over the past 700 kyr in the subantarctic Pacific. The
UK’37-derived SST records display varying glacial severity, as opposed to the more uniform
relative glacial/interglacial change in the UK37-inferred SST records. Based on the better
agreement of the glacial severity patterns of the UK37 records with that of the planktic 18O at
the same sites and other subantarctic SST records, we conclude that the UK37 is a more
suitable index for paleo SST reconstruction in the subantarctic Pacific. The UK37-derived SST
records suggest pronounced glacial cooling of ~8°C and ~4°C in the subantarctic and the
subtropical regions, respectively. The magnitude of subantarctic glacial cooling is comparable
to that reported for other sectors of the Southern Ocean. The SST estimates also suggest that
the ACC and its associated fronts migrated equatorward by 7° to 9° during glacials over the
past 700 kyr, which might have reduced the water transport through the Drake Passage to the
South Atlantic. Conversely, the deflection of more ACC waters equatorward during glacials
probably enhanced the cold water advection via the PCC, resulting in colder subtropical SSTs
and, thus, larger meridional SST gradients between the tropics and the subtropics.
3.7. Acknowledgements
Constructive comments from three anonymous reviewers and the editor Rainer Zahn
have improved the manuscript tremendously. We are grateful to Giuseppe Cortese for sharing
his unpublished data. Thanks also go to Gemma Rueda, Susanne Fietz, Nicole Meyer and
Hendrik Grotheer for assisting in the alkenone analysis. Technical support from Walter
Luttmer and Ralph Kreutz in the laboratories is also highly appreciated. We acknowledge the
POLMAR Graduate School for funding S.L.H’s PhD study at the Alfred Wegener Institute
and a SCAR Fellowship 2010/2011 for funding S.L.H’s short stay at the Autonomous
University of Barcelona during the preparation of PS75 core-top samples.
71 Chapter 4 Chapter 4: Diverging Pleistocene SST trends revealed by UK37 and TEX86H
in the Southeast Pacific
Sze Ling Ho1*, Gesine Mollenhauer1, Frank Lamy1, Mahyar Mohtadi2, Ralf Tiedemann1
1
Alfred Wegener Institute for Polar and Marine Research, P.O.Box 12 01 61, 27515 Bremerhaven, Germany.
[*Corresponding author: [email protected]]
2
MARUM-Center for Marine Environmental Sciences, University of Bremen, Leobener Strasse, D-28359
Bremen, Germany.
(Manuscript in preparation)
4.0. Abstract
Multi-proxy approach is essential for constraining sea surface temperature (SST)
reconstruction. Organic SST proxies based on the alkenones (of haptophyte) and the archaeal
glycerol dialkyl glycerol tetraether (GDGT) are useful in this regard. Here we present a new
GDGT-inferred SST record and compare it to a previously published alkenone-derived SST
record from the same site. The comparison reveals different temporal patterns and values in
both records. The GDGT-based SST estimates are generally colder. To explain the differences
between these proxies, we assess both seasonality and habitat depth of the source organisms,
and conclude that the latter is a more likely reason. Based on previous findings of maximum
archaeal abundance at the oxycline in this region, we hypothesize that the colder GDGTinferred estimates reflect the water temperature at this hydrographic boundary instead of the
sea surface, as in the alkenone record. The only exception to this pattern occurs during MIS
11 and 13, when the GDGT-based temperatures are considerably warmer than other
interglacials, leading to a convergence of temperature values inferred from both proxies. This
phenomenon is probably attributable to variability in the water column structure in the
Southeast Pacific. One plausible mechanism is a more drastic deepening in the thermocline
relative to the oxycline. In this scenario, the oxycline corresponds to the warmer upper part of
the thermocline instead of the base of thermocline as observed in the modern day
climatological data.
72 Chapter 4 4.1. Introduction
Given the importance of the ocean in climate system, knowledge of past variability in
sea surface temperature (SST) is essential for understanding the mechanism involved in
climate changes and for validating the numerical models that are widely used for projection of
future climate. A large array of proxies have been developed to reconstruct past SST changes,
such as the foraminiferal 18O and Mg/Ca, the alkenone unsaturation index and the transfer
functions of various microfossils. Very often these proxies result in dissimilar reconstructed
SST patterns, due to differences in the sensitivity of their response to environmental change,
the habitat depth and the production seasonality of the source organisms, or other secondary
non-thermal effects. Therefore, it is important to constrain SST reconstructions with multiple
proxies. In this regard, organic geochemical proxies based on alkenones and glycerol dialkyl
glycerol tetraethers (GDGT) are good options for carrying out a multi-proxy approach, since
they can be extracted simultaneously from marine sediments, rendering this approach less
labor intensive than comparing organic-inorganic proxies. The alkenones unsaturation index,
i.e., the UK37 is based on the lipids of haptophyte algae, including the cosmopolitan
coccolithophores Emiliania huxleyi [Brassell et al., 1986; Prahl and Wakeham, 1987]. The
GDGT-based proxy, termed TetraEther index with 86 carbon atoms (abbreviated as TEX86)
[Schouten et al., 2002], is based on the membrane lipids of mesophilic marine archaea from
group Thaumarchaeota (formerly known as Marine Group 1 Crenarchaeota). Recent
development has seen increasing numbers of publications based on this organic multi-proxy
approach, including studies in the Arabian Sea [Huguet et al., 2006b], the Mediterranean
[Castañeda et al., 2010; Huguet et al., 2011], the equatorial Atlantic [Lopes dos Santos et al.,
2010], the Gulf of California [McClymont et al., 2012] and the Agulhas system [Caley et al.,
2011]. While some of these studies showed similar SST patterns derived from both proxies
[e.g., Castañeda et al., 2010], others not [Huguet et al., 2006b; Lopes dos Santos et al., 2010].
These authors attributed the differences between alkenone and GDGT proxies to several
possibilities, e.g., the GDGTs record subsurface temperature instead of SST [e.g., Lopes dos
Santos et al., 2010; McClymont et al., 2012], or these lipids are produced in different seasons,
thus are not controlled by the same climatic systems [Huguet et al., 2006b]. Given that the
isoprenoid GDGTs used in TEX86 are also found in soil, it is important to assess whether the
GDGT-derived SST estimates are biased by terrestrial input. Hopmans et al. [2004] proposed
an index known as Branched and Isoprenoid Tetraether (BIT) index as a qualitative parameter
to gauge the influence of terrestrial end-member on the sedimentary GDGT pool. A threshold
73 Chapter 4 value of 0.3 (above which the SST estimates might be biased) was suggested by Weijers et al.
[2006]. However, recent studies [e.g., Fietz et al., 2011; Smith et al., 2012] questioned its
usefulness in tracing the changes in fluvial input as the ratio could be overwhelmed by the
marine end member.
Here we present a GDGT-inferred SST record over the past 700 kyr in the relatively
under-studied Southeast Pacific, and compare it with a previously published alkenone SST
record at the same study site [Ho et al., in press]. We use BIT values and the concentration of
GDGTs to evaluate potential bias introduced by terrestrial GDGTs on the GDGT-inferred
SST estimates, before discussing the patterns in both SST records.
4.2. Study site
Core GeoB 3388-1 (25°13’S 75°31’W, 3558 m water depth) was retrieved from the
Iquique Ridge approximately 500 km off Chilean coast during the R/V Sonne cruise 102
[Hebbeln et al., 1995]. The site lies underneath a vigorous eastern boundary current, i.e., the
oceanic Peru-Chile Current (PCC, also known as the Humboldt Current). The PCC carries
cold, nutrient-rich water originates from the Antarctic Circumpolar Current (ACC)
equatorward, feeding into the South Equatorial Current (SEC). The nutrients upwelled in the
coastal PCC zone result in high productivity, which in turn leads to high consumption of
dissolved oxygen for organic matter decomposition. This high oxygen demand, along with
sluggish ventilation in the region, produce a subsurface oxygen minimum zone (OMZ; <0.5
mL/L oxygen concentration) [Helly and Levin, 2004] as thick as 500 m [Fuenzalida et al.,
2009]. The upper boundary of the OMZ deepens and the intensity diminishes further offshore
(Figure 4.1). The thickness, the position and the intensity of the OMZ could be influenced by
the El Niño Southern Oscillation (ENSO) [Morales et al., 1999; reference therein]. The depth
of the upper boundary of the OMZ (referred to as the oxycline in the text), is generally deeper
than the thermocline (see also Figure 4.6), and both deepen considerably during the warm
ENSO phase, i.e., the El Niño. The coastal OMZ water is advected poleward by the PeruChile Undercurrent (PUC, also known as the Poleward Undercurrent and the Gunther
Current). The PUC is the poleward branch formed after the bifurcation of the Equatorial
Undercurrent (EUC) as it approaches the Galapagos Islands[Strub et al., 1998].
74 Chapter 4 Figure 4.1: Core locations of GeoB 3388-1 and TG7 [Calvo et al., 2001], and major hydrographic features in the
Southeast Pacific. Black full line denotes surface currents while black dashed line denotes subsurface currents.
Right panel illustrate the oxygen concentration along a section at 25°S between the Chilean coast and subtropical
gyre (illustrated by the blue dotted line in the left panel). Abbreviations: ACC – Antarctic Circumpolar Current;
PCC- Peru-Chile Current; SEC – South Equatorial Current; EUC – Equatorial Undercurrent Current; PUC –
Peru-Chile Undercurrent.
4.3. Materials and methods
Piston core GeoB 3388-1 was previously described by Mohtadi et al. [2006] and Ho et
al. [in press]. Here we opt for the age model revised by Ho et al. [in press], which is based on
the visual tuning of the benthic 18O record to the global benthic 18O stack of Lisiecki and
Raymo [2005]. According to this age model, the average sedimentation rate at the site is 0.7
cm/kyr. Lipids were extracted from freeze-dried, homogenized sediments, via sonication with
successively less polar solvent mixture (dichloromethane and methanol), as described by
Müller et al. [1997]. Total lipid extracts were separated into two fractions (alkenones and
GDGTs) using column chromatography (silica cartridge) prior to analysis of individual
compounds. GDGTs were eluted using methanol and passed through PTFE filters before they
were analyzed using a High Performance Liquid Chromatography system coupled to a mass
spectrometer with an atmospheric pressure chemical ionization interface (HPLC-APCI-MS).
The setting of the instrumental condition was modified from the method of Hopmans et al.
75 Chapter 4 [2000] as reported previously by Ho et al. [submitted to GCA]. GDGTs were quantified using
internal standard, i.e., C46 GDGT.
SST estimates were calculated using a modified version of the TEX86, known as
H
TEX86 , following Kim et al. [2010]:
H
TEX86
Log
GDGT‐2 GDGT‐3 Cren'
GDGT‐1 GDGT‐2 GDGT‐3 Cren'
SST 68.4
H
46.9
TEX 86
where Cren’ denotes crenarchaeol regio-isomer.
We also calculated the BIT index values to assess the relative abundance of marine to soil
GDGTs, as proposed by Hopmans et al. [2004]:
BIT
GDGT I GDGT II GDGT III
GDGT I GDGT II GDGT III Crenarchaeol
Figure 4.2: Molecular structures of glycerol dialkyl glycerol tetraethers (GDGTs) used for the calculation of the
TEX86H and BIT indices [Castañeda et al., 2010].
76 Chapter 4 4.4. Results and discussion
2
24
4
6
Marine Isotope Stage (MIS)
8
10
12
14
16
(a)
16
12
Crenarchaeol
(b)
Branched GDGTs
1.0
(c)
0.6
0.4
3.0
(d)
3.5
4.0
0
50
100
150
200
250
300
350
400
450
500
550
600
650
4.5
700
0
0.0
18
2.5
0.2
Benthic  O ( /00)
BIT
0.8
40
35
30
25
20
15
10
5
0
GDGT conc. (ng/g sed)
SST (°C)
20
Age (kyr)
Figure 4.3: Glycerol dialkyl glycerol tetraether (GDGT) parameters and benthic 18O record of core GeoB 33881. Blue shaded bars represent glacial intervals while the black numbers on top denote the marine isotope stages
(a) TEX86H-derived SST estimates computed using the global core-top calibration of Kim et al. [2010]. The blue
and red circles denote the present day WOA09 coldest and warmest months, respectively. (b) Concentration of
branched GDGTs and creanarchaeol used in the calculation of the BIT values. (c) Branched and Isoprenoid
Tetraether (BIT) index values. (d) Benthic 18O values of Cibicidoides wuellerstorfi previously published by
Mohtadi et al. [2006].
The TEX86H-derived SST record displays glacial-interglacial (G-IG) fluctuations with
values ranging between 22°C and 10°C over the past 700 kyr at site GeoB 3388-1. The record
is rather noisy with some drastic jumps especially at the lower part of the record. There also
seems to be some G-IG temporal offsets between the TEX86H record with the foraminiferal
benthic 18O values, where warm peaks appear in glacial intervals, e.g., MIS 6 and MIS 14.
77 Chapter 4 TEX86H suggests Holocene warmth (~9 kyr) of around 20.7°C (the top of the piston core) that
is comparable to modern day warmest month SST (20.2°C from WOA09). The average
glacial SST values are colder than the present day winter SST.
4.4.1. High BIT values: potential terrestrial bias on TEX86H SST?
Before making any climatic interpretation of the TEX86H-derived record, it is
necessary to assess whether the record has been compromised by terrestrial GDGTs, in light
of the high BIT values (>0.3) at our study site for most part of the past 700 kyr. The BIT
index was initially proposed as a proxy to monitor the fluvial input to marine sediment.
Weijers et al. [2006] found that soil GDGTs caused an artificial warming in TEX86-derived
SSTs and suggested a cut-off value of 0.3, above which the temperature estimates might
become unreliable. The BIT values at our site are in the range of 0 and 0.8, with a decreasing
trend towards the Holocene. The trend in the BIT record is clearly governed by the variations
in the crenarchaeol concentrations, as the concentrations of branched GDGTs are invariant
throughout the record (Figure 4.3). Given the considerable distance from land (~500km) and
the absence of major rivers on land around the latitude of our core site, substantial fluvial
input of branched GDGTs to site GeoB 3388-1 is unlikely. Furthermore, the little fluvial
sediments that end up in the ocean are probably caught in the Peru-Chile Trench, which lies
between the coast and our study site. Previous sedimentological evidences suggest that the
terrigenous components west of the trench are eolian-derived [Lamy et al., 1998; Saukel et al.,
2011]. To-date it is still unclear whether the GDGTs could be transported by winds. However,
if indeed such massive amount of dust-entrained GDGTs are blown to our study site and
interfere with the marine GDGT signal, one would expect to see an overall “warm” TEX86Hderived SST throughout the record consistent with the invariant concentration of branched
GDGTs. Warm bias on TEX86-derived temperature estimates by terrestrial GDGTs is
suggested by Weijers et al. [2006] and the TEX86H-inferred temperature values (approx. 29°C)
in three soil samples from the Atacama Desert at subtropical latitudes (see Section 4.7). Since
the interglacial TEX86H SSTs at site GeoB 3388-1 are comparable with modern day SSTs and
not anomalously warm, it is likely that a terrigenous overprint on this record is negligible.
Furthermore, there is no apparent correlation between the BIT index values and the SST
estimates (Figure 4.4). The branched GDGTs in our sediment record are probably produced
in situ, e.g., in the water column or the sediment by bacteria, as previously invoked by Peterse
78 Chapter 4 et al. [2009] and Zhu et al. [2011] to explain the occurrence of branched GDGTs in the
marine sediments off Svalbard and the South China Sea, respectively. High relative
abundances of branched GDGTs (hence BIT values) have also been observed in marine
sediments from the Southern Ocean [Fietz et al., 2011] and the equatorial Atlantic [Fietz et
al., 2012] far from any apparent fluvial sources (e.g., major rivers), further support the
plausibility of in-situ production of branched GDGTs in open ocean settings. Since the
branched GDGTs-producing bacteria are not known to produce isoprenoid GDGTs, we
contend that high BIT values do not compromise the validity of the TEX86H-inferred SST
estimates at our study site.
Figure 4.4: Correlation between the Branched and Isoprenoid Tetraether (BIT) index values and the TEX86Hderived SST estimates calculated with the global core-top calibration of Kim et al. [2010].
4.4.2. Comparison with alkenone-derived SSTs
In spite of lower temporal resolutions, the alkenone-inferred SST record at site GeoB
3388-1 resembles the subtropical TG7 alkenone SST record [Calvo et al., 2001] which lies a
few degrees northward (see Figure 4.1 for location), suggesting that both records likely
reflect the regional SST pattern. Comparison of the alkenone-derived and the smoothed
TEX86H-derived SST records suggests that for most part of the past 700 kyr, the G-IG
amplitude of SST oscillation is ~4°C. The only exception was during MIS 11 and 13, where
the TEX86H suggests larger amplitudes of up to 8°C. In general, the alkenone-derived SST
estimates are warmer by approximately 5°C than those inferred from the TEX86H, except for
MIS 11 and 13 where the estimates inferred from both proxies converge. While there is no
79 Chapter 4 clear G-IG trends in the differences between the alkenone SST and the TEX86H SST (termed
SST), the SST during MIS 11 and 13 are evidently reduced compared to other time
intervals. This begs the questions – what causes the differences between the alkenone- and the
TEX86H-derived SSTs? Why do they converge during MIS 11 and 13?
2
24
4
6
Marine Isotope Stage (MIS)
8
10
12
TG7
(a)
14
16
GeoB 3388-1 (Alkenone)
16
12
H
GeoB 3388-1 (TEX86 )
-5
(b)
0
5
10
0
50
100
150
200
250
300
350
400
Age (kyr)
450
500
550
600
650
700
Alkenone - TEX86H SST (°C)
SST (°C)
20
Figure 4.5: Temperature records at site GeoB 3388-1. Black lines denote records smoothed by a running average
method. Blue shaded bars represent glacial intervals while the black numbers on top denote the marine isotope
stages. (a) Alkenone-derived sea surface temperature records at site GeoB 3388-1 (blue) and TG7 (red) [Calvo et
al., 2001], and TEX86H-derived water temperature record (green) at site GeoB 3388-1. The blue and red circles
denote the present day WOA09 coldest and warmest months, respectively, at site GeoB 3388-1. (b) The
difference between alkenone-derived and TEX86H-derived temperature estimates at site GeoB 3388-1.
Firstly, we explore the plausible reasons for the differences between the two proxies.
Taking the SST estimates at face value and assuming that the condition of previous
interglacials are comparable to that of present day, one might speculate that the alkenones and
the GDGTs are produced during summer and winter, respectively, judging by their agreement
with the respective WOA09 seasonal SSTs (see Figure 4.5). Nonetheless, this idea cannot be
better constrained in the absence of recent sediment sample at the study site (the top of core
GeoB 3388-1 is estimated to be 9000 year-old) and sediment trap time series data in this
region. A regional core-top alkenone study in the Southeast Pacific indeed suggests that the
80 Chapter 4 sedimentary alkenones in this region register annual mean SST [Prahl et al., 2006], as
opposed to our summer-biased alkenone-inferred interglacial SSTs
To the best of our knowledge, there is no reported work on the abundance and
seasonal variability of GDGTs in the water column or surface sediment in the vicinity of our
study site. Meanwhile, time series data on the prokaryote in the PCC system reveal no
seasonal patterns in the abundance of marine archaea (precursors of GDGTs) [Quiñones et al.,
2009]. We note that studies from other regions suggested that the Thaumarchaeota thrive
during winter in the coastal North Sea [Wuchter et al., 2005]) and the Antarctic coastal waters
[Church et al., 2003; Murray et al., 1998]. However, we refrain from extrapolating these
findings obtained from studies on shallow coastal waters to our open ocean site, which is
situated in a completely different hydrographic and sedimentation setting. This is in light of a
recent finding by Leider et al. [2010] in a core-top study in the Mediterranean Sea. They
attributed the dissimilar TEX86H-SST relationships observed at coastal sites and oceanic sites
to differences in the seasonality and/or sedimentation setting. Furthermore, Giovannoni and
Vergin [2012] also proposed that the microbial communities differ between coastal and open
ocean regions, and with latitude.
Alternatively, dissimilar SST values can also arise if the alkenones and the GDGTs
record water temperatures at different depths. Unlike the photosynthetic precursor of the
alkenones (i.e., haptophyte algae) which are found in the upper photic zone, the
chemoautotrophic archaea thrive throughout the water column [e.g., Karner et al., 2001] since
their habitat depth is not restricted by light. Off northern Chile, maximum absolute abundance
of Thaumarchaeotal cell numbers were found at the oxycline (water depths of ~50 – 200m)
above the core of the OMZ [Belmar et al., 2011]. If indeed the GDGTs in the sediment
originate from this water depth which coincides with the base of thermocline, they should
record water temperatures about 6°C lower than the SST (Figure 4.6), in agreement with the
SST observed in the alkenone- and TEX86H records. GDGT-based proxies reflecting
subsurface temperatures have been reported previously, such as for the Gulf of California
[McClymont et al., 2012], the eastern tropical Atlantic [Lopes dos Santos et al., 2010] and the
Benguela upwelling system [Lee et al., 2008]. We find this notion more plausible that the
seasonality bias, thus adopt it as our working hypothesis for further discussion.
81 Chapter 4 4.4.3. Paleoceanographic implications
Assuming that the alkenones and the GDGTs record water temperatures at the sea
surface and the oxycline, respectively, these proxy data suggest that the water column
condition is anomalous during MIS 11 and 13, where the oxycline temperatures warm up
twice as much as the other time intervals and are almost as warm as the SST. The warming is
likely to be restricted to the subsurface since the alkenone-derived MIS 11 and 13 SSTs are
not anomalously warm compared to other interglacials. If the archaeal preferred habitat depth
remains the same during these intervals, i.e., at the oxycline, significant subsurface warming
during MIS 11 and 13 suggests an enhanced influence of warm waters at the oxycline. At
present, the oxycline at site GeoB 3388-1 corresponds to the base of the thermocline. As the
temperatures at the base of thermocline do not vary much with depth variation (Note: the
thermocline is theoretically the boundary between two water masses with specific physical
properties), same magnitude of depth shifts in both the oxycline and the thermocline will not
lead to large changes in the TEX86H-derived oxycline temperature and in turn the SST.
Conversely, substantial oxycline temperature change requires a mismatch in the magnitude of
shoaling / deepening of both these hydrographic boundaries.
The variability of the oxycline depth, i.e., the upper boundary of the OMZ, in the
subtropical Southeast Pacific is closely linked to the changes in thermocline depth and
upwelling intensity [Fuenzalida et al., 2009], which are in turn closely associated with the
ENSO variability. During warm El Niño events, large parcel of warm waters accumulates in
the eastern Pacific and the upwelling of cold, nutrient rich water at the coastal area is reduced.
The depths of both thermocline and oxycline deepen; and the opposite are true for La Niña
condition. Morales et al. [1999] reported that during an El Niño event offhore Antofagasta
(23°S), the thermocline deepened more than did the oxycline. Consequently, the oxycline
corresponds to the upper part of the thermocline, unlike during “normal” conditions when the
oxycline corresponds to the base of the thermocline. Considering that it is warmer at the upper
part of the thermocline than at the base of it, such an El Niño-like reorganization of water
column structure would result in smaller temperature difference between the sea surface and
the oxycline (Figure 4.6). Due to the thicker upper warm water mass, the upwelled water will
be warmer than usual, further contribute to decrease the surface-subsurface temperature
difference. Putting our temperature reconstruction in this context and assuming that the signal
82 Chapter 4 depths of alkenone and GDGTs are constant through time, smaller SSTs suggest a change in
the water column structure analogous to that occurring during the El Niño events.
El Niño-like MIS 11 and 13 in the Southeast Pacific has been previously proposed by
Mohtadi et al. [2006] based on an independent proxy on core GeoB 3388-1, i.e., foraminiferal
13C values. The authors invoked a reorganization of the PCC circulation pattern in the
Southeast Pacific to explain the diverging planktonic 13C values in the tropical and the
subtropical Southeast Pacific during MIS 11 and 13, i.e., the weakened PCC turned westward
at a much southerly position than usual leading to this divergence. As argued by the authors,
such a shift in the circulation system resembles the El Niño conditions, where weakened
Hadley Cell and Walker Circulation lead to a weak gyre circulation and a decreased influence
of the ACC (via the PCC) in the equatorial East Pacific. Our water column temperature
reconstruction further suggests that during these time intervals, the thermocline deepened
more than the oxycline compared to other time intervals over the past 700 kyr.
Figure 4.6: Upper water column structure at site GeoB 3388-1 in terms of oxygen concentration and temperature
based on the annual mean climatological data of WOA09 (green curves). The red curves are the schematic of the
water column structure during El Niño events. The horizontal dashed lines illustrate the position of oxycline and
its corresponding temperatures. The arrows depict the difference between the temperatures at the sea surface and
the oxycline where the Thaumarchaeota thrive.
83 Chapter 4 4.5. Conclusions
In spite of high BIT index values, it is unlikely that our TEX86H record in the open
ocean subtropical Southeast Pacific has been compromised by terrestrial GDGTs because the
SST estimates are not anomalously warm and they show no correlation with the BIT index
values. The branched GDGTs are probably produced in-situ in the water column or the
sediment as proposed for other oceanic regions by previous workers [Peterse et al., 2009; Zhu
et al., 2011]. The TEX86H-derived SST estimates are generally 5°C colder than those inferred
from the alkenones, probably because the GDGTs record water temperature at the oxycline
while the alkenones the surface. This notion is supported by microbial phylogenetic study in
the region which found the highest absolute abundance of marine Thaumarchaeota at the
oxycline above the oxygen minimum zone [Belmar et al., 2011]. This pattern breaks down
during MIS 11 and 13, during which SST estimates derived from both proxies converge. We
invoke a drastically deepened thermocline (relative to the oxycline) to explain the substantial
warming recorded by the TEX86H during these time intervals. Such a shift resembles the
changes in the water column during the El Niño events in modern day. El Niño-like MIS 11
and MIS 13 was previously suggested by Mohtadi et al. [2006] based on the foraminiferal
13C data measured on the same core.
4.6. Acknowledgments
We thank Carme Huguet and Susanne Fietz for their encouragement and discussion on
earlier results. Technical assistance of Ralph Kreutz (handling the HPLC-APCI-MS) is
greatly appreciated. We also thank Dierk Hebbeln and Claudio Latorre for providing sediment
samples of core GeoB 3388-1 and soil samples from the Atacama Desert, respectively. The
POLMAR Graduate School is acknowledged for funding S.L.H.’s PhD study at the Alfred
Wegener Institute.
4.7. Supplementary Information
Branched and Isoprenoid Tetraether (BIT) index values following the equation of Hopmans et al.
[2004], temperature estimates calculated with the TEX86H calibration of Kim et al. [2010], and
coordinates of the soil samples from the Atacama Desert.
84 Chapter 4 Site
BIT
TEX86H T
Longitude
Latitude
Altitude
(°W)
(°S)
(m)
L01
67.72
23.50
4480
0.85
28.3
L11
67.85
23.32
3570
0.61
27.3
L16
67.92
23.31
3070
0.72
30.7
(°C)
85 Chapter 5 Chapter 5: Conclusions and Outlook
5.1. Summary and conclusions
This thesis contributes to a better understanding of organic sea surface temperature
(SST) proxies derived from alkenones and archaeal glycerol dialkyl glycerol tetraethers
(GDGTs) in three aspects, i.e., the spatial distribution of these lipids, the calibrations and the
indices of the proxies, and the resulting paleoclimatic implications.
5.1.1. Lipids occurrence: Where the proxies may be applied
Alkenones are found between the Subtropical Front and the Antarctic Polar Front
(APF) in the South Pacific (Figure 3.1), with highest abundances found in the shallow waters
off New Zealand. Application of alkenone paleothermometry in the cold waters south of the
APF and further south in the Antarctic zone would therefore require large amounts of
sediment samples to have alkenones in quantifiable abundance. The GDGTs are found in
subpolar and polar regions, including all sectors of the Southern Ocean, the North Pacific, the
Bering Sea, the Arctic Ocean and the Fram Strait (Figure 2.1). The presence of GDGTs at
high latitudes renders them promising for SST reconstruction at sites where the absence of the
alkenones precludes the application of the better established SST indices e.g., the alkenonebased UK’37.
5.1.2. Proxy indices and calibrations: Diverging views from core-top calibration and
downcore application
The correlations between the two alkenone unsaturation indices (UK37 and UK’37) in the
South Pacific surface sediments and the annual mean World Ocean Atlas 2009 (WOA09)
climatological SSTs (1°C - 12°C) are identical to the commonly used laboratory E. huxleyi
culture-based calibrations of Prahl et al. [1988] (8°C – 25°C) (Figure 3.3). The South Pacific
alkenone indices – SST correlations are linear, with equally high r2 values of ~0.9 for both
indices. These findings contradict previous studies that suggest a loss of linearity in the
alkenone indices – SST relationships at low temperatures [Conte et al., 2006; Sikes and
86 Chapter 5 Volkman, 1993; Sikes et al., 1997], and that the UK’37-SST correlation (r2 = 0.9) is better than
that of UK37-SST (r2 = 0.7) [Sikes et al., 1997]. Although the South Pacific core-top data
support the use of both alkenone indices for SST reconstruction in the subantarctic region,
downcore results suggest otherwise. The two alkenone indices yield diverging Pleistocene
SST patterns for the study sites in the subantarctic Southern Ocean (Figure 3.4). Judging
from the better structural fit with planktonic foraminiferal 18O records at these sites and other
subantarctic SST records derived from various proxies (Figure 3.5), it appears that the
original UK37 index (instead of the commonly used simplified UK’37 index) results in more
plausible Pleistocene SST records, hence is a more suitable alkenone SST index in the
subantarctic Southern Ocean.
The 160 GDGT indices (TEX86 and TEX86L) data presented in this work are discussed
in combination with ~470 previously published data to obtain the largest spatial coverage to
date (Figure 2.1). The correlations of the index values in subpolar and polar core-tops with
WOA09 annual mean SST are poor (r2 values < 0.3) (Figure 2.4). Cross-plotting these
indices data with seasonal WOA09 SST data sets, and water temperature at various
hydrographic boundaries (seasonal thermocline, oxycline, deepest water temperature for each
site) does not result in improved correlations (Figure 2.5 and Figure 2.6), suggesting that the
scatter in the data is caused by other factors, such as differences in sedimentation setting.
Running linear regressions through the new global compilation of core-top GDGT indices
data (n = 630) does not yield significantly different equations compared to the previous global
calibrations proposed by Kim et al. [2010] (Figure 2.9). The results from the new global
compilation show that the correlation for TEX86-SST is better than that for the TEX86L-SST,
and the TEX86 index results in better SST estimates for subpolar and polar regions (Figure
2.7), in contrast to the conclusions of a previous global core-top calibration study [Kim et al.,
2010]. Meanwhile, the new Arctic core-top data presented here show that both GDGT indices
overestimate the SST here by up to 30°C (relative to the annual mean SST), especially in the
vicinity of the sea ice margin. These Arctic data aside, the scatter at the low temperature end
is not more pronounced than that in the mid temperature range, suggesting that the previously
suggested loss of temperature dependence at low temperatures might not be true. Therefore,
the newly compiled global data set does not suggest the need for different GDGT indices for
application at opposite ends of the SST range.
87 Chapter 5 Although the core-top data indicate that GDGT indices correlate better with annual
mean SST instead of seasonal SST or subsurface temperatures, a downcore TEX86H
(logarithmic version of the TEX86 proposed by Kim et al. [2010]) record suggests that this
might not be the case in the subtropical South Pacific. In this 700 kyr sediment record, the
GDGT-derived SSTs are colder than those inferred from the alkenones. The interglacial
estimates are colder than annual mean WOA09, and more comparable with modern-day
winter SST at the site (Figure 4.5). Considering the lack of seasonality in the archaeal
abundance in the subtropical South Pacific [Quiñones et al., 2009], it seems that the “coldbiased” GDGT-inferred SSTs here reflect subsurface temperature, plausibly at the oxycline
because the source organism, i.e., the Thaumarchaeota, thrive at this hydrographic boundary
[Belmar et al., 2011].
5.1.3. Paleo SST estimates: Pleistocene climatic implications
Two subantarctic and one subtropical SST records presented in this thesis shed light
on the SST variability in the Southeast Pacific over the past 700 kyr. The subantarctic records
are especially important in terms of the long timescale they cover, which is roughly ten times
older than the longest SST record in the region. They afford a high-to-low latitudes SST
patterns comparison based on records derived from the same proxy (i.e., UK37), which is
essential for constraining latitudinal SST gradient calculation. The UK37 SST records show
that in the Southeast Pacific over the past 700 kyr, the glacial cooling in the subantarctic zone
(~8°C) is twice larger than that in the subtropics (~4°C) (Figure 3.4), resulting in larger
latitudinal SST gradients during glacials (Figure 3.6). The extent of glacial cooling here is
comparable, if not larger, than what observed in other sectors of the subantarctic Southern
Ocean (Figure 3.5), in contrast to the findings of Gersonde et al. [2005] based on siliceous
microfossil transfer functions, which suggest less intense glacial cooling in the Pacific sector.
Although sharing a first-order pattern with the Antarctic temperature evolution registered in
ice cores [Jouzel et al., 2007; Petit et al., 1999], the subantarctic Pacific SST records do not
display a well-expressed Mid-Brunhes Event as in the ice core record. These UK37-derived
SSTs imply massive equatorward migrations (by 7 to 9 degrees latitude) of the Antarctic
Circumpolar Current frontal systems and enhanced cold water transport to the low latitudes
via the Peru-Chile Current (PCC) during glacials.
88 Chapter 5 Whilst the alkenone-inferred interglacial warmth in the subtropics appears to be quite
constant over the past 700 kyr, the GDGT-derived temperatures suggest a two-fold subsurface
warming at the oxycline during MIS 11 and 13 (Figure 4.5). Assuming that the present-day
lack of seasonality in the archaeal abundance holds in the past (hence ruling out a shift in
archaeal production season), the substantial subsurface warming might be related to an
enhanced deepening of the thermocline relative to the oxycline, causing the latter (also the
habitat depth of the archaea) to correspond to the upper, warmer part of the thermocline
(Figure 4.6). Such shifts in the depths of these hydrographic boundaries are analogous to the
water column reorganization during modern-day El Niño conditions [Morales et al., 1999].
5.2. Perspectives for future work
“Science is always wrong: It never solves a problem without creating ten more.”
George Bernard Shaw (1856-1950)
As solemnly stated in a community White Paper [Henderson et al., 2009] prepared for
the National Environment Research Council of the United Kingdom following a symposium
on the earth’s climate convened by Henry Elderfield and colleagues at the University of
Cambridge, “Despite its importance, careful calibration of a paleoproxy and assessment of its
uncertainty can appear less exciting to a researcher or a funding panel than the application
of this proxy to a paleoclimate question”. There is clearly room for improvement in proxy
calibration and its application.
As discussed thoroughly in previous chapters, choosing the “right” index and
calibration is pivotal for reconstructing past SST changes. Indeed, it has been shown that
different approaches in terrestrial temperature calibrations alone can result in differences of
the same magnitude as the temperature changes in the past millennium reported in the IPCC
report [Esper et al., 2005]. In alkenone- and GDGT-inferred SST records, the temporal trends
are governed by the definition of the index while the values of the SST estimates depend on
the calibration employed. It goes without saying that both the index and the calibration of SST
proxies are of paramount importance for accurately reconstructing past SST changes. In light
89 Chapter 5 of the findings presented in this thesis, many more scientific questions arise, which are useful
for steering future work to further improve the quality of SST reconstruction.
5.2.1. Refining UK37 calibration
Chapter 3 presents South Pacific alkenone core-top calibrations that are comparable
with the commonly used Emiliania huxleyi culture calibrations [Prahl et al., 1988], but differ
systematically from the Southern Ocean core-top calibrations of Sikes et al. [1997], especially
evident for the UK37 index (Figure 5.1). Inconsistency in instrumentation and methodology,
including chromatogram integration and sample work-up, could be possible causative factors
for this offset [personal communication with E. Sikes]. Notwithstanding, given the rarity of
the Southern Ocean core-top samples, it is worthy to further examine other potential factors
causing the offset, such as non-thermal physiological factors, differences in assemblages or
morphotypes in alkenone-producing coccolithophores, before ruling out one of the data sets.
This is especially true considering the difficulty in culturing coccolithophores for calibration
at temperatures below 5°C. In addition to low growth rates (resulting in a time-consuming
experiment), the cold water coccolithophores cultures isolated from the Southern Ocean
waters have a small range of preferred growth temperature (~2°C) [G. Langer, personal
communication] that is not ideal for calibrating the SST proxy.
Furthermore, in light of the new finding presented in this thesis, i.e., the UK37 index is
a better SST proxy in the subantarctic region at low temperatures (Section 3.5.1 and Section
3.5.2), there is an urgent need to appraise the index at higher temperatures to expand its
calibration range. One possible short-term approach would be to combine the South Pacific
core-top data with the E. huxleyi culture data of Prahl et al. [1988], yielding a calibration
range from 1 °C to 25°C (compared to 0 °C to 28°C for the global core-top UK’37 calibration).
Such approach was adopted by Mashiotta et al. [1999] for their “hybrid” foraminiferal Mg/Ca
calibration, which closely resembles the core-top calibration of Elderfield and Ganssen
[2000]. Notwithstanding, it is not ideal in theory to combine temporally integrated signals in
surface sediments with “snapshot” signals in laboratory cultures. Therefore, the long-term
goal should be to generate more UK37 data spanning the entire global temperature range,
which will not only help to reconcile the differences in various data sets (Figure 5.1), but will
ultimately contribute to generating a UK37 global calibration with a comparable temperature
90 Chapter 5 range to its counterpart UK’37. A universal calibration for the entire temperature range would
enable a more quantitative comparison of SST records spanning a large latitudinal range.
1.2
1
0.8
UK37
0.6
0.4
0.2
0
‐0.2
‐0.4
0
5
10
15
20
25
30
Temperature (°C)
PS75
Rosell‐Mele95
Linear (Sikes97)
Prahl88
Linear (PS75)
Sikes97
Linear (Prahl88)
Figure 5.1: UK37 calibrations available in the literature. Prahl et al. [1988]’s calibration is based on the
Emiliania huxleyi cultures. Calibrations of Rosell-Mele et al. [1995] and Sikes et al. [1997] are based on core-top
data from the North Atlantic and the Southern Ocean, respectively.
Wide acceptance of the UK37 index as SST proxy can be promoted by further evidence
that within the same type of alkenone producers (coccolithophore), variations in the relative
abundance of C37:4 alkenones (termed %C37:4 herein) in the open ocean are not caused by
changes in salinity. %C37:4 as a salinity proxy was proposed by Rosell-Melé [1998] and
Rosell-Melé et al. [2002] based on core-top data from the North Atlantic and the Nordic Seas.
However, as argued by Sikes and Sicre [2002], the proposed relationship could be an artifact
due to the strong correlation between temperature and salinity, which incidentally also holds
in the South Pacific (Figure 5.2). Furthermore, in coastal area, fjords, brackish waters and
sea-ice margins, the variations in %C37:4 are likely due to changes in the alkenone producer
assemblages [Bendle et al., 2005; Marlowe et al., 1984; Schulz et al., 2000], not because of
cellular physiological response to salinity variations. Considering the difficulty to disentangle
the effects of these co-varying factors in the natural environments (i.e., shift in assemblages
and strong correlation between temperature and salinity), laboratory culture experiments (e.g.,
growing coccolithophores at constant temperature while changing salinity) will provide more
definitive insights into the effect of salinity on %C37:4 in coccolithophores. This might have
important implications for future application of alkenones as a SST proxy and/or an indicator
for low salinity water masses. Some studies have used the UK’37 index in parallel with the
91 Chapter 5 %C37:4 to infer SSTs and the presence of low salinity water, respectively. Although this
approach is convenient and has its merits given that a salinity proxy is still elusive, it is
lacking a biogeochemical rationale. That is, if salinity does control the alkenone unsaturation
pattern (thereby also the %C37:4), it would violate the principle of the UK’37 index (which is
practically an index that quantifies the degree of unsaturation) as a SST proxy. Alternatively,
if the correlation between %C37:4 and salinity stems from the contribution of coastal/brackish
water/fresh water alkenone producers which have different alkenone distributions than their
open ocean counterparts (i.e., with higher abundance of C37:4 alkenones), these alkenones will
“contaminate” the signals of marine alkenones, rendering the SST estimates derived from
these “mixed alkenones” using the marine core-top calibrations inaccurate. In any case,
parallel application of alkenones as both SST and salinity proxy cannot be justified.
34.6
SSS (psu)
34.4
34.2
34
R² = 0.9333
33.8
33.6
0
2
4
6
8
SST (°C)
10
12
14
Figure 5.2: Correlation between the sea surface salinity (SSS) and the sea surface temperature (SST) at the coretop sites of PS75 reported in Chapter 3. Environmental data are retrieved from the WOA09.
5.2.2. Refining GDGT-based SST calibrations
Compared to the extensively studied alkenone paleothermometry, there is still plenty
of room for improvement in GDGT-based proxies due to their more recent discovery (16
years later than the alkenone paleothermometry) and their relatively under-studied archaeal
precursors. It is seen in Chapter 2 that the addition of 230 data points to the global core-top
calibration data sets [Kim et al., 2010] does not lead to very different regressions. Instead, the
additional data result in larger scatter and hence, larger errors of estimation. This finding
suggests that the more urgent issue for GDGT indices is to reduce these errors by
investigating the causative factors for the scatter as speculated based on core-top data in
92 Chapter 5 Chapter 2. Such factors include the seasonality in the archaeal abundance and their export to
the sediments, potential GDGT contribution from an archaeal group other than the mesophilic
Thaumarchaeota, and the archaeal GDGTs originated from subsurface water depths instead of
the sea surface.
To test the various hypotheses listed above, sediment trap time series from different
hydrographic settings have the potential to unravel regional differences, if any, in the
seasonality and the sinking of the GDGT fluxes. To-date, most sediment trap data in the
literature are from the tropics and/or shallow continental shelves [Fallet et al., 2011;
Wakeham et al., 2003; Wuchter et al., 2006]. Therefore, sediment trap data from open ocean
settings and higher latitudes may provide new insights into these issues.
Nonetheless, sediment trap sites worldwide are currently, and will likely be in the
future, insufficient to match the broad spatial coverage of the global core-top data set. As an
alternative to the sediment trap data, one can probably make use of satellite observations on
primary productivity that have global coverage (1° x 1° grid). In addition, the export of
primary production from the sea surface can be predicted using biogeochemical models, albeit
with some extent of uncertainties inherent to the models. Furthermore, the temperature
residuals of TEX86L (Figure 5.3) display latitudinal patterns, where overestimates
(underestimates) are observed at high (low) latitudes. These patterns resemble the seasonality
index calculated by Schneider et al. [2010], which shows that primary productivity is
positively (negatively) correlated to SST at high (low) latitudes. Therefore, incorporating
biogeochemical parameters in TEX86 calibrations has the potential to improve the calibration,
such as changing the slope of the regression (Figure 5.4). Conventionally, seasonality in
global calibrations is examined by comparing the r2 values in the crossplots of the SST index
values and seasonal SSTs. However, the seasonal patterns in production and export of organic
matter are not comparable at low and high latitudes, with maxima occurring in different
seasons. Therefore, a strong justification for correlating index values with seasonal SSTs in
global calibrations is lacking.
More confidence in the TEX86 paleothermometry will ensue if a better constraint on
the original export depth of the GDGTs and more knowledge on the GDGT-producing
archaea can be obtained. The fact that the alleged GDGT source organisms (the
Thaumarchaeota) thrive throughout the water column casts some doubts on the validity of
GDGTs as a SST proxy. The present assumption is that GDGTs are transported to the
93 Chapter 5 sediment via scavenging and fecal pellet formation, thus they reflect the temperatures at upper
water depths where an active foodweb exists. Future study is needed to further corroborate
this assumption. In this regard, microbial phylogenetic analysis performed in tandem with
lipid analysis on suspended matter in the water column at various depths might be useful to
improve our understanding of the source organisms and their preferred habitat depth. Such
combined analyses on suspended matters in the water column off Chile are currently
underway in the laboratory of O. Ulloa, the University of Concepcion in collaboration with R.
Summons (Massachusetts Institute of Technology) [O. Ulloa, personal communication]. In
°C
Figure 5.3: TEX86L-derived temperature residuals (difference between estimated SST and climatological SST)
by applying the current global calibration of Kim et al. [2010] on published data sets [Ho et al., 2011; Kim et al.,
2010; Leider et al., 2010; Shevenell et al., 2011] and data presented in Chapter 2.
Figure 5.4: Schematic of a possible shift in the global core-top calibrations of TEX86 (currently calibrated
against annual mean SST) after integrating production and export information, which skew towards warm (cold)
seasons at high (low) latitudes.
94 Chapter 5 addition, knowledge on the correlation between the abundances of the archaea and the
GDGTs will provide some perspectives on how to translate the currently available archaeal
abundance data into the distribution of GDGTs in the water column. Laboratory culture
experiment would be ideal for studying GDGT-producing archaea, but culturing mesophilic
archaea has proven to be very challenging so far, and there is yet no reported success in
culturing the Thaumarchaeota.
5.2.3. Constraining paleo SST reconstruction: Multi-proxy approach
As summarized in Section 5.1.2, diverging views on the proxy applicability could be
inferred from core-top data and downcore records, suggesting that the modern proxy-SST
relationship in the core-top calibration might falter in the past. In this regard, a multi-proxy
approach can help to better constrain SST reconstruction. When the proxies are in agreement,
they provide additional confidence in the SST estimates. Conversely, disagreements in the
proxies might reveal potential bias and prompt us to find the underlying processes to reconcile
the differences.
For instance, glacial SSTs at site PS75/034-2 are ~1°C (Figure 3.4). At present, such
low temperatures are found south of the APF near the winter sea ice edge, where
coccolithophores are sparse or absent [Cubillos et al., 2007; Gravalosa et al., 2008; Hasle,
1960] and alkenones are at detection limit (note however that alkenones have been found in
sediments where the overlying SST is ~0°C in the South Atlantic [Müller et al., 1998]).
Notwithstanding, the alkenone abundances in core PS75/034-2 are higher during glacials than
interglacials. This suggests that either the alkenones production / preservation is substantially
higher during glacials, or the modern alkenone calibration underestimates the glacial SSTs in
the past. Additional oceanographic information provided by other SST proxy and sea ice
proxy (e.g., diatom assemblages) will be useful to determine which hypothesis is more likely.
Multi-proxy comparison is especially crucial for advancing our understanding of the
predictive power of the TEX86 index in estimating past temperature changes, considering the
uncertainties associated with the depth of origin for GDGTs. In light of sediment trap studies
that suggest a subsurface origin for GDGTs [Huguet et al., 2007; Lee et al., 2008], one can
examine whether this is true for the past by comparing the TEX86 records with other SST
proxies and subsurface temperature proxies (e.g., Mg/Ca on deep-dwelling foraminifera such
95 Chapter 5 as G. inflata and G. truncatulinoides) records measured on the same sediment cores. More
importantly, such comparison will also yield insights into whether the source depth for
GDGTs has shifted in the past. This has implications for the assumption made (that the
GDGT signal depth remains invariant in the past) in the hypothesis put forward in Chapter 4
to explain the substantial warming registered in the GDGT-derived temperature record during
MIS 11 and 13.
5.2.4. Towards the “big picture”
As stated in Chapter 1, the ultimate goal of all branches of climate research is to have
an accurate projection of future climate change, in order to better prepare for its consequences
on human’s society. Since proxy data are widely used to validate numerical models,
improvements of both proxies and models are pivotal towards achieving that ultimate goal.
Recent development has seen heightened awareness in the scientific community regarding the
importance of bringing together paleoclimatology and modeling. Further examination on
proxy calibrations and the quantification of their uncertainties have the potential to contribute
to this cause. Some studies have shown that the temperature variations simulated by climate
models are smaller than those observed in proxy data [Braconnot et al., 2012; Lohmann et al.,
2012; Lorenz et al., 2006; Schneider et al., 2010]. As proposed in Section 5.2.2, integrating
biogeochemical parameters such as production or export might lead to a shift in the slope of
the global calibration, resulting in smaller amplitudes of the SST variation for the same range
of index values. Therefore, future work shall go beyond the conventional scope of paleo
proxy work and attempt to place proxy studies in the context of model-data comparison. The
advantages of model-proxy comparison are analogous to those of the multi-proxy comparison,
i.e., agreement shall bring confidence in both models and proxy data; while disagreements
encourage the community to strive harder to understand underlying causes of the mismatch.
Furthermore, changes in global calibrations shall have far-reaching implications for the
calculations of climate sensitivity over glacial-interglacial cycles, which in turn affect the
projection of future climate with a doubling of atmospheric CO2.
96 References References
Alonso-Sáez, L., Sánchez, O., Gasol, J.M., Balagué, V., Pedrós-Alio, C., 2008. Winter-tosummer changes in the composition and single-cell activity of near-surface Arctic
prokaryotes. Environmental Microbiology 10, 2444-2454.
Bano, N., Ruffin, S., Ransom, B., Hollibaugh, J.T., 2004. Phylogenetic composition of Arctic
Ocean archaeal assemblages and comparison with Antarctic assemblages. Applied and
Environmental Microbiology 70, 781-789.
Bard, E., 2001. Comparison of alkenone estimates with other paleotemperature proxies.
Geochemistry Geophysics Geosystems 2, 1002.
Bard, E., Rickaby, R.E.M., 2009. Migration of the subtropical front as a modulator of glacial
climate. Nature 460, 380-383.
Barker, S., Cacho, I., Benway, H., Tachikawa, K., 2005. Planktonic foraminiferal Mg/Ca as a
proxy for past oceanic temperatures: a methodological overview and data compilation for the
Last Glacial Maximum. Quaternary Science Reviews 24, 821-834.
Barrows, T.T., Juggins, S., 2005. Sea-surface temperatures around the Australian margin and
Indian Ocean during the Last Glacial Maximum. Quaternary Science Reviews 24, 1017-1047.
Beck, J.W., Edwards, R.L., Ito, E., Taylor, F.W., Recy, J., Rougerie, F., Joannot, P., Henin,
C., 1992. Sea-Surface Temperature from Coral Skeletal Strontium/Calcium Ratios. Science
257, 644-647.
Becquey, S., Gersonde, R., 2002. Past hydrographic and climatic changes in the Subantarctic
Zone of the South Atlantic - The Pleistocene record from ODP Site 1090. Palaeogeography,
Palaeoclimatology, Palaeoecology 182, 221-239.
Becquey, S., Gersonde, R., 2003. A 0.55-Ma paleotemperature record from the Subantarctic
zone: Implications for Antarctic Circumpolar Current development. Paleoceanography 18,
1014.
Bell, M.V., Pond, D., 1996. Lipid composition during growth of motile and coccolith forms of
Emiliania huxleyi. Phytochemistry 41, 465-471.
Belmar, L., Molina, V., Ulloa, O., 2011. Abundance and phylogenetic identity of
archaeoplankton in the permanent oxygen minimum zone of the eastern tropical South
Pacific. FEMS Microbiology Ecology 78, 314-326.
Bendle, J., Rosell-Melé, A., 2004. Distributions of UK37 and UK'37 in the surface waters and
sediments of the Nordic Seas: Implications for paleoceanography. Geochemistry Geophysics
Geosystems 5, Q11013.
Bendle, J., Rosell-Melé, A., Ziveri, P., 2005. Variability of unusual distributions of alkenones
in the surface waters of the Nordic seas. Paleoceanography 20, PA2001.
97 References Benthien, A., Müller, P.J., 2000. Anomalously low alkenone temperatures caused by lateral
particle and sediment transport in the Malvinas Current region, western Argentine Basin.
Deep-Sea Research Part I-Oceanographic Research Papers 47, 2369-2393.
Blaga, C.I., Reichart, G.J., Heiri, O., Damste, J.S.S., 2009. Tetraether membrane lipid
distributions in water-column particulate matter and sediments: a study of 47 European lakes
along a north-south transect. Journal of Paleolimnology 41, 523-540.
Braconnot, P., Harrison, S.P., Kageyama, M., Bartlein, P.J., Masson-Delmotte, V., AbeOuchi, A., Otto-Bliesner, B., Zhao, Y., 2012. Evaluation of climate models using
palaeoclimatic data. Nature Climate Change 2, 417-424.
Brassell, S.C., Eglinton, G., Marlowe, I.T., Pflaumann, U., Sarnthein, M., 1986. Molecular
stratigraphy - a new tool for climatic assessment. Nature 320, 129-133.
Brochier-Armanet, C., Boussau, B., Gribaldo, S., Forterre, P., 2008. Mesophilic
crenarchaeota: proposal for a third archaeal phylum, the Thaumarchaeota. Nature Reviews
Microbiology 6, 245-252.
Caley, T., Kim, J.-H., Malaize, B., Giraudeau, J., Laepple, T., Caillon, N., Charlier, K.,
Rebaubier, H., Rossignol, L., Castaneda, I.S., Schouten, S., Sinninghe Damste, J.S., 2011.
High-latitude obliquity as a dominant forcing in the Agulhas current system. Climate of the
Past 7, 1285 -1296.
Calvo, E., Grimalt, J.O., Jansen, E., 2002. High resolution UK 37 sea surface temperature
reconstruction in the Norwegian Sea during the Holocene. Quaternary Science Reviews 21,
1385-1394.
Calvo, E., Pelejero, C., Herguera, J.C., Palanques, A., Grimalt, J.O., 2001. Insolation
dependence of the southeastern Subtropical Pacific sea surface temperature over the last 400
kyrs. Geophysical Research Letters 28, 2481-2484.
Caniupán, M., Lamy, F., Lange, C.B., Kaiser, J., Arz, H., Kilian, R., Baeza Urrea, O.,
Aracena, C., Hebbeln, D., Kissel, C., Laj, C., Mollenhauer, G., Tiedemann, R., 2011.
Millennial-scale sea surface temperature and Patagonian Ice Sheet changes off southernmost
Chile (53°S) over the past 60 kyr. Paleoceanography 26, PA3221.
Casey, K.S., Cornillon, P., 1999. A comparison of satellite and in situ based sea surface
temperature climatologies. Journal of Climate 12, 3801-3818.
Castañeda, I.S., Schefuß, E., Pätzold, J., Sinninghe Damsté, J.S., Weldeab, S., Schouten, S.,
2010. Millennial-scale sea surface temperature changes in the eastern Mediterranean (Nile
River Delta region) over the last 27,000 years. Paleoceanography 25, PA1208.
Chaigneau, A., Pizarro, O., 2005. Surface circulation and fronts of the South Pacific Ocean,
east of 120°W. Geophysical Research Letters 32, L08605.
Chave, K.E., 1954. Aspects of the biogeochemistry of magnesium: 1. Calcareous marine
organisms. Journal of Geology, 266-283.
98 References Chavez, F.P., Messié, M., 2009. A comparison of Eastern Boundary Upwelling Ecosystems.
Progress In Oceanography 83, 80-96.
Church, M.J., DeLong, E.F., Ducklow, H.W., Karner, M.B., Preston, C.M., Karl, D.M., 2003.
Abundance and distribution of planktonic Archaea and Bacteria in the waters west of the
Antarctic Peninsula. Limnolog and Oceanography 48, 1893-1902.
Conte, M.H., Eglinton, G., 1993. Alkenone and Alkenoate Distributions within the Euphotic
Zone of the Eastern North-Atlantic - Correlation with Production Temperature. Deep-Sea
Research Part I- Oceanographic Research Pappers 40, 1935-1961.
Conte, M.H., Sicre, M.A., Ruhlemann, C., Weber, J.C., Schulte, S., Schulz-Bull, D., Blanz,
T., 2006. Global temperature calibration of the alkenone unsaturation index (UK'37) in surface
waters and comparison with surface sediments. Geochemistry Geophysics Geosystems 7,
Q02005.
Cortese, G., Abelmann, A., 2002. Radiolarian-based paleotemperatures during the last 160
kyr at ODP Site 1089 (Southern Ocean, Atlantic Sector). Palaeogeography,
Palaeoclimatology, Palaeoecology 182, 259-286.
Crosta, X., Sturm, A., Armand, L., Pichon, J.-J., 2004. Late Quaternary sea ice history in the
Indian sector of the Southern Ocean as recorded by diatom assemblages. Marine
Micropaleontology 50, 209-223.
Cubillos, J.C., Wright, S.W., Nash, G., de Salas, M.F., Griffiths, B., Tilbrook, B., Poisson, A.,
Hallegraeff, G.M., 2007. Calcification morphotypes of the coccolithophorid Emiliania huxleyi
in the Southern Ocean: changes in 2001 to 2006 compared to historical data. Marine Ecology
Progress Series 348, 47-54.
Cunningham, S.A., Alderson, S.G., King, B.A., Brandon, M.A., 2003. Transport and
variability of the Antarctic Circumpolar Current in Drake Passage. Journal of Geophysical
Research 108, 8084.
de Leeuw, J.W., v.d. Meer, F.W., Rijpstra, W.I.C., Schenck, P.A., 1980. On the occurrence
and structural identification of long chain unsaturated ketones and hydrocarbons in sediments.
Physics and Chemistry of The Earth 12, 211-217.
De Rosa, M., Gambacorta, A., 1988. The lipids of archaebacteria. Progress in Lipid Research
27, 153-175.
de Vernal, A., Rosell-Melé, A., Kucera, M., Hillaire-Marcel, C., Eynaud, F., Weinelt, M.,
Dokken, T., Kageyama, M., 2006. Comparing proxies for the reconstruction of LGM seasurface conditions in the northern North Atlantic. Quaternary Science Reviews 25, 28202834.
Deser, C., Alexander, M.A., Xie, S.-P., Phillips, A.S., 2010. Sea Surface Temperature
Variability: Patterns and Mechanisms. Annual Review of Marine Science 2, 115-143.
Elderfield, H., Ganssen, G., 2000. Past temperature and 18O of surface ocean waters inferred
from foraminiferal Mg/Ca ratios. Nature 405, 442-445.
99 References Emiliani, C., 1955. Pleistocene temperatures. Journal of Geology 63, 538-578.
Epstein, B.L., D'Hondt, S., Hargraves, P.E., 2001. The possible metabolic role of C-37
alkenones in Emiliania huxleyi. Organic Geochemistry 32, 867-875.
Epstein, S., Buchsbaum, R., Lowenstam, H.A., Urey, H.C., 1953. Revised carbonate-water
isotopic temperature scale. Geological Society of America Bulletin 64, 1315-1326.
Escala, M., Rosell-Melé, A., Masque, P., 2007. Rapid screening of glycerol dialkyl glycerol
tetraethers in continental Eurasia samples using HPLC/APCI-ion trap mass spectrometry.
Organic Geochemistry 38, 161-164.
Esper, J., Frank, D.C., Wilson, R.J.S., Briffa, K.R., 2005. Effect of scaling and regression on
reconstructed temperature amplitude for the past millennium. Geophysical Research Letters
32, L07711.
Fallet, U., Ullgren, J.E., Castañeda, I.S., van Aken, H.M., Schouten, S., Ridderinkhof, H.,
Brummer, G.-J.A., 2011. Contrasting variability in foraminiferal and organic
paleotemperature proxies in sedimenting particles of the Mozambique Channel (SW Indian
Ocean). Geochimica et Cosmochimica Acta 75, 5834-5848.
Feldberg, M.J., Mix, A.C., 2002. Sea-surface temperature estimates in the Southeast Pacific
based on planktonic foraminiferal species; modern calibration and Last Glacial Maximum.
Marine Micropaleontology 44, 1-29.
Feldberg, M.J., Mix, A.C., 2003. Planktonic foraminifera, sea surface temperatures, and
mechanisms of oceanic change in the Peru and south equatorial currents, 0-150 ka BP.
Paleoceanography 18, 1016.
Fietz, S., Huguet, C., Bendle, J., Escala, M., Gallacher, C., Herfort, L., Jamieson, R.,
Martínez-Garcia, A., McClymont, E.L., Peck, V.L., Prahl, F.G., Rossi, S., Rueda, G., SansonBarrera, A., Rosell-Melé, A., 2012. Co-variation of crenarchaeol and branched GDGTs in
globally-distributed marine and freshwater sedimentary archives. Global and Planetary
Change 92–93, 275-285.
Fietz, S., Martínez-Garcia, A., Huguet, C., Rueda, G., Rosell-Melé, A., 2011. Constraints in
the application of the Branched and Isoprenoid Tetraether index as a terrestrial input proxy.
Journal of Geophysical Research-Oceans 116, C10032.
Fischer, H., Schmitt, J., Lüthi, D., Stocker, T.F., Tschumi, T., Parekh, P., Joos, F., Köhler, P.,
Völker, C., Gersonde, R., Barbante, C., Le Floch, M., Raynaud, D., Wolff, E., 2010. The role
of Southern Ocean processes in orbital and millennial CO2 variations – A synthesis.
Quaternary Science Reviews 29, 193-205.
Fuenzalida, R., Schneider, W., Garcés-Vargas, J., Bravo, L., Lange, C., 2009. Vertical and
horizontal extension of the oxygen minimum zone in the eastern South Pacific Ocean. Deep
Sea Research Part II: Topical Studies in Oceanography 56, 992-1003.
100 References Gagan, M.K., Ayliffe, L.K., Beck, J.W., Cole, J.E., Druffel, E.R.M., Dunbar, R.B., Schrag,
D.P., 2000. New views of tropical paleoclimates from corals. Quaternary Science Reviews 19,
45-64.
Gaines, S.M., Eglinton, G., Rullkotter, J., 2008. Echoes of Life: What Fossil Molecules
Reveal about Earth History. Oxford University Press.
Gersonde, R., 2011. The expedition of the research vessel "Polarstern" to the polar South
Pacific in 2009/2010 (ANT-XXVI/2 - BIPOMAC), In: Gersonde, R. (Ed.), Berichte zur Polarund Meeresforschung (Reports on Polar and Marine Research). Alfred Wegener Institute for
Polar and Marine Research, Bremerhaven, p. 330.
Gersonde, R., Abelmann, A., Brathauer, U., Becquey, S., Bianchi, C., Cortese, G., Grobe, H.,
Kuhn, G., Niebler, H.S., Segl, M., Sieger, R., Zielinski, U., Futterer, D.K., 2003. Last glacial
sea surface temperatures and sea-ice extent in the Southern Ocean (Atlantic-Indian sector): A
multiproxy approach. Paleoceanography 18.
Gersonde, R., Crosta, X., Abelmann, A., Armand, L., 2005. Sea-surface temperature and sea
ice distribution of the Southern Ocean at the EPILOG Last Glacial Maximum - A circumAntarctic view based on siliceous microfossil records. Quaternary Science Reviews 24, 869896.
Giovannoni, S.J., Vergin, K.L., 2012. Seasonality in Ocean Microbial Communities. Science
335, 671-676.
Gliozzi, A., Paoli, G., De Rosa, M., Gambacorta, A., 1983. Effect of isoprenoid cyclization on
the transition temperature of lipids in thermophilic archaebacteria. Biochimica et Biophysica
Acta - Biomembranes 735, 234-242.
Gliozzi, A., Relini, A., Chong, P.L.G., 2002. Structure and permeability properties of
biomimetic membranes of bolaform archaeal tetraether lipids. Journal of Membrane Science
206, 131-147.
Gravalosa, J.M., Flores, J.A., Sierro, F.J., Gersonde, R., 2008. Sea surface distribution of
coccolithophores in the eastern Pacific sector of the Southern Ocean (Bellingshausen and
Amundsen Seas) during the late austral summer of 2001. Marine Micropaleontology 69, 1625.
Harada, N., Ahagon, N., Sakamoto, T., Uchida, M., Ikehara, M., Shibata, Y., 2006. Rapid
fluctuation of alkenone temperature in the southwestern Okhotsk Sea during the past 120 ky.
Global and Planetary Change 53, 29-46.
Hasle, G.R., 1960. Plankton coccolithophorids from the Subantarctic and Equatorial Pacific.
Nytt Magasin for Botanikk 8, 77-88.
Hastie, T., Tibshirani, R., 1990. Generalized Additive Models. Chapman & Hall, London.
Hays, J.D., Imbrie, J., Shackleton, N.J., 1976. Variations in the earth's orbit: Pacemaker of the
ice ages. Science 194, 1121-1132.
101 References Hazel, J.R., 1988. Homeoviscous adaptation in animal cell membranes, Physiological
Regulation of Membrane Fluidity, pp. 149-188.
Hebbeln, D., Wefer, G., Participants, C., 1995. Cruise report of R/V Sonne cruise 102,
Valparaiso-Valparaiso,, Rep. 68. Univ. Bremen, Bremen, p. 166.
Helly, J.J., Levin, L.A., 2004. Global distribution of naturally occurring marine hypoxia on
continental margins. Deep-Sea Res. Pt. I 51, 1159-1168.
Henderson, G., Collins, M., Hall, I., Lockwood, M., Palike, H., Rickaby, R., Schmidt, G.,
Turney, C., Wolff, E., 2009. Improving future climate prediction using palaeoclimate data, pp.
1-28.
Hendy, E.J., Gagan, M.K., Alibert, C.A., McCulloch, M.T., Lough, J.M., Isdale, P.J., 2002.
Abrupt Decrease in Tropical Pacific Sea Surface Salinity at End of Little Ice Age. Science
295, 1511-1514.
Herbert, T.D., 2003. Alkenone Paleotemperature Determinations, In: Holland, H.D.,
Turekian, K.K. (Eds.), Treatise on Geochemistry. Pergamon, Oxford, pp. 391-432.
Ho, S.L., Mollenhauer, G., Lamy, F., Martínez-Garcia, A., Mohtadi, M., Gersonde, R.,
Hebbeln, D., Nunez Ricardo, S., Rosell-Melé, A., Tiedemann, R., in press. Sea surface
temperature variability in the Pacific sector of the Southern Ocean over the past 700 kyr.
Paleoceanography.
Ho, S.L., Yamamoto, M., Mollenhauer, G., Minagawa, M., 2011. Core top TEX86 values in
the south and equatorial Pacific. Organic Geochemistry 42, 94-99.
Honda, M.C., Imai, K., Nojiri, Y., Hoshi, F., Sugawara, T., Kusakabe, M., 2002. The
biological pump in the northwestern North Pacific based on fluxes and major components of
particulate matter obtained by sediment-trap experiments (1997-2000). Deep-Sea Research
Part II-Topical Studies in Oceanography 49, 5595-5625.
Honjo, S., Francois, R., Manganini, S., Dymond, J., Collier, R., 2000. Particle fluxes to the
interior of the Southern Ocean in the Western Pacific sector along 170 degrees W. Deep-Sea
Research Part II-Topical Studies in Oceanography 47, 3521-3548.
Hopmans, E.C., Schouten, S., Pancost, R.D., van der Meer, M.T.J., Sinninghe Damsté, J.S.,
2000. Analysis of intact tetraether lipids in archaeal cell material and sediments by high
performance liquid chromatography/atmospheric pressure chemical ionization mass
spectrometry. Rapid Communication in Mass Spectrometry 14, 585-589.
Hopmans, E.C., Weijers, J.W.H., Schefuss, E., Herfort, L., Sinninghe Damsté, J.S., Schouten,
S., 2004. A novel proxy for terrestrial organic matter in sediments based on branched and
isoprenoid tetraether lipids. Earth and Planetary Science Letters 224, 107-116.
Horikawa, K., Murayama, M., Minagawa, M., Kato, Y., Sagawa, T., 2010. Latitudinal and
downcore (0–750 ka) changes in n-alkane chain lengths in the eastern equatorial Pacific.
Quaternary Research 73, 573-582.
102 References Howard, W.R., Prell, W.L., 1992. Late Quaternary Surface Circulation of the Southern Indian
Ocean and its Relationship to Orbital Variations. Paleoceanography 7, 79-117.
Huguet, C., Cartes, J.E., Sinninghe Damsté, J.S., Schouten, S., 2006a. Marine crenarchaeotal
membrane lipids in decapods: Implications for the TEX86 paleothermometer. Geochemistry
Geophysics Geosystems 7, Q11010.
Huguet, C., Kim, J.H., Sinninghe Damsté, J.S., Schouten, S., 2006b. Reconstruction of sea
surface temperature variations in the Arabian Sea over the last 23 kyr using organic proxies
(TEX86 and UK'37). Paleoceanography 21, PA3003.
Huguet, C., Martrat, B., Grimalt, J.O., Sinninghe Damsté, J.S., Schouten, S., 2011. Coherent
millennial-scale patterns in UK'37 and TEXH86 temperature records during the penultimate
interglacial-to-glacial cycle in the western Mediterranean. Paleoceanography 26, PA2218.
Huguet, C., Schimmelmann, A., Thunell, R., Lourens, L.J., Sinninghe Damsté, J.S., Schouten,
S., 2007. A study of the TEX86 paleothermometer in the water column and sediments of the
Santa Barbara Basin, California. Paleoceanography 22, PA3203.
Huybers, P., Wunsch, C., 2010. Paleophysical Oceanography with an Emphasis on Transport
Rates. Annual Review of Marine Science 2, 1-34.
Imbrie, J., Kipp, N.G., 1977. A new micropaleontological method for Quantitative
Paleoclimatology: Application to a late PleistoceneCaribbean Core., In: Turekian, K.K. (Ed.),
The Late Cenozoic Glacial Ages. Yale University Press, New Haven, pp. 71-181.
Jouzel, J., Masson-Delmotte, V., Cattani, O., Dreyfus, G., Falourd, S., Hoffmann, G., Minster,
B., Nouet, J., Barnola, J.M., Chappellaz, J., Fischer, H., Gallet, J.C., Johnsen, S.,
Leuenberger, M., Loulergue, L., Luethi, D., Oerter, H., Parrenin, F., Raisbeck, G., Raynaud,
D., Schilt, A., Schwander, J., Selmo, E., Souchez, R., Spahni, R., Stauffer, B., Steffensen,
J.P., Stenni, B., Stocker, T.F., Tison, J.L., Werner, M., Wolff, E.W., 2007. Orbital and
millennial Antarctic climate variability over the past 800,000 years. Science 317, 793-796.
Junge, K., Eicken, H., Deming, J.W., 2004. Bacterial Activity at −2 to −20°C in Arctic
Wintertime Sea Ice. Applied and Environmental Microbiology 70, 550-557.
Kaiser, J., Lamy, F., Hebbeln, D., 2005. A 70-kyr sea surface temperature record off southern
Chile (Ocean Drilling Program Site 1233). Paleoceanography 20, PA4009.
Kalanetra, K.M., Bano, N., Hollibaugh, J.T., 2009. Ammonia-oxidizing Archaea in the Arctic
Ocean and Antarctic coastal waters. Environ Microbiol 11, 2434-2445.
Karner, M.B., DeLong, E.F., Karl, D.M., 2001. Archaeal dominance in the mesopelagic zone
of the Pacific Ocean. Nature 409, 507-510.
Kidston, J., Taschetto, A.S., Thompson, D.W.J., England, M.H., 2011. The influence of
Southern Hemisphere sea-ice extent on the latitude of the mid-latitude jet stream. Geophysical
Research Letters 38, L15804.
103 References Kienast, M., MacIntyre, G., Dubois, N., Higginson, S., Normandeau, C., Chazen, C., Herbert,
T.D., 2012. Alkenone unsaturation in surface sediments from the eastern equatorial Pacific:
Implications for SST reconstructions. Paleoceanography 27, PA1210.
Kim, J.H., Schneider, R.R., Hebbeln, D., Müller, P.J., Wefer, G., 2002. Last deglacial seasurface temperature evolution in the Southeast Pacific compared to climate changes on the
South American continent. Quaternary Science Reviews 21, 2085-2097.
Kim, J.H., Schouten, S., Hopmans, E.C., Donner, B., Sinninghe Damsté, J.S., 2008. Global
sediment core-top calibration of the TEX86 paleothermometer in the ocean. Geochimica et
Cosmochimica Acta 72, 1154-1173.
Kim, J.H., van der Meer, J., Schouten, S., Helmke, P., Willmott, V., Sangiorgi, F., Koc, N.,
Hopmans, E.C., Sinninghe Damsté, J.S., 2010. New indices and calibrations derived from the
distribution of crenarchaeal isoprenoid tetraether lipids: Implications for past sea surface
temperature reconstructions. Geochimica et Cosmochimica Acta 74, 4639-4654.
Könneke, M., Bernhard, A.E., de la Torre, J.R., Walker, C.B., Waterbury, J.B., Stahl, D.A.,
2005. Isolation of an autotrophic ammonia-oxidizing marine archaeon. Nature 437, 543-546.
Kucera, M., Rosell-Melé, A., Schneider, R., Waelbroeck, C., Weinelt, M., 2005. Multiproxy
approach for the reconstruction of the glacial ocean surface (MARGO). Quaternary Science
Reviews 24, 813-819.
Lamy, F., Hebbeln, D., Wefer, G., 1998. Terrigenous sediment supply along the Chilean
continental margin: modern regional patterns of texture and composition. Geologische
Rundschau 87, 477-494.
Lamy, F., Kaiser, J., Ninnemann, U., Hebbeln, D., Arz, H.W., Stoner, J., 2004. Antarctic
Timing of Surface Water Changes off Chile and Patagonian Ice Sheet Response. Science 304,
1959-1962.
Lang, N., Wolff, E.W., 2011. Interglacial and glacial variability from the last 800 ka in
marine, ice and terrestrial archives. Climate of the Past 7, 361-380.
Lee, K.E., Kim, J.H., Wilke, I., Helmke, P., Schouten, S., 2008. A study of the alkenone,
TEX86, and planktonic foraminifera in the Benguela Upwelling System: Implications for past
sea surface temperature estimates. Geochemistry Geophysics Geosystems 9, Q10019.
Leider, A., Hinrichs, K.-U., Mollenhauer, G., Versteegh, G.J.M., 2010. Core-top calibration
of the lipid-based and TEX86 temperature proxies on the southern Italian shelf (SW Adriatic
Sea, Gulf of Taranto). Earth and Planetary Science Letters 300, 112-124.
Lipp, J.S., Hinrichs, K.U., 2009. Structural diversity and fate of intact polar lipids in marine
sediments. Geochimica et Cosmochimica Acta 73, 6816-6833.
Lisiecki, L.E., Raymo, M.E., 2005. A Pliocene-Pleistocene stack of 57 globally distributed
benthic 18O records. Paleoceanography 20, PA1003.
104 References Liu, Z.H., Pagani, M., Zinniker, D., DeConto, R., Huber, M., Brinkhuis, H., Shah, S.R.,
Leckie, R.M., Pearson, A., 2009. Global Cooling During the Eocene-Oligocene Climate
Transition. Science 323, 1187-1190.
Locarnini, R.A., Mishonov, A.V., Antonov, J.I., Boyer, T.P., Garcia, H.E., Baranova, O.K.,
Zweng, M.M., Johnson, D.R., 2010. World Ocean Atlas 2009: Volume 1: Temperature, In:
Levitus, S. (Ed.), NOAA Atlas NESDIS 68. U.S. Government Printing Office, Washington,
D.C., p. 184.
Lohmann, G., Pfeiffer, M., Laepple, T., Leduc, G., Kim, J.H., 2012. A model-data
comparison of the Holocene global sea surface temperature evolution. Climate of the Past
Discussion 8, 1005-1056.
Lopes dos Santos, R.A., Prange, M., Castañeda, I.S., Schefuß, E., Mulitza, S., Schulz, M.,
Niedermeyer, E.M., Sinninghe Damsté, J.S., Schouten, S., 2010. Glacial–interglacial
variability in Atlantic meridional overturning circulation and thermocline adjustments in the
tropical North Atlantic. Earth and Planetary Science Letters 300, 407-414.
Lorenz, S.J., Kim, J.-H., Rimbu, N., Schneider, R.R., Lohmann, G., 2006. Orbitally driven
insolation forcing on Holocene climate trends: Evidence from alkenone data and climate
modeling. Paleoceanography 21, PA1002.
Luz, B., 1977. Late Pleistocene paleoclimates of the South Pacific based on statistical analysis
of planktonic foraminifers. Palaeogeography, Palaeoclimatology, Palaeoecology 22, 61-78.
Malmgren, B.A., Kucera, M., Nyberg, J., Waelbroeck, C., 2001. Comparison of Statistical
and Artificial Neural Network Techniques for Estimating Past Sea Surface Temperatures from
Planktonic Foraminifer Census Data. Paleoceanography 16, 520-530.
Margot, H., Acebal, C., Toril, E., Amils, R., Fernandez Puentes, J., 2002. Consistent
association of crenarchaeal Archaea with sponges of the genus Axinella. Marine Biology 140,
739-745.
Marlowe, I.T., Brassell, S.C., Eglinton, G., Green, J.C., 1990. Long-Chain Alkenones and
Alkyl Alkenoates and the Fossil Coccolith Record of Marine-Sediments. Chemical Geology
88, 349-375.
Marlowe, I.T., Green, J.C., Neal, A.C., Brassell, S.C., Eglinton, G., Course, P.A., 1984. Long
chain (n-C37–C39) alkenones in the Prymnesiophyceae. Distribution of alkenones and other
lipids and their taxonomic significance. British Phycological Journal 19, 203-216.
Marshall, J., Speer, K., 2012. Closure of the meridional overturning circulation through
Southern Ocean upwelling. Nature Geoscience 5, 171-180.
Martínez-Garcia, A., Rosell-Melé, A., Geibert, W., Gersonde, R., Masque, P., Gaspari, V.,
Barbante, C., 2009. Links between iron supply, marine productivity, sea surface temperature,
and CO2 over the last 1.1 Ma. Paleoceanography 24, PA1207.
105 References Martínez-Garcia, A., Rosell-Melé, A., McClymont, E.L., Gersonde, R., Haug, G.H., 2010.
Subpolar link to the emergence of the modern equatorial Pacific cold tongue. Science 328,
1550-1553.
Mashiotta, T.A., Lea, D.W., Spero, H.J., 1999. Glacial-interglacial changes in Subantarctic
sea surface temperature and delta O-18-water using foraminiferal Mg. Earth and Planetary
Science Letters 170, 417-432.
McClymont, E.L., Ganeshram, R.S., Pichevin, L.E., Talbot, H.M., van Dongen, B.E., Thunell,
R.C., Haywood, A.M., Singarayer, J.S., Valdes, P.J., 2012. Sea-surface temperature records of
Termination 1 in the Gulf of California: Challenges for seasonal and interannual analogues of
tropical Pacific climate change. Paleoceanography 27, PA2202.
McClymont, E.L., Martinez-Garcia, A., Rosell-Mele, A., 2007. Benefits of freeze-drying
sediments for the analysis of total chlorins and alkenone concentrations in marine sediments.
Organic Geochemistry 38, 1002-1007.
Mohtadi, M., Hebbeln, D., 2004. Mechanisms and variations of the paleoproductivity off
northern Chile (24°S & 33°S) during the last 40,000 years. Paleoceanography 19, PA2023.
Mohtadi, M., Hebbeln, D., Nuñez Ricardo, S., Lange, C.B., 2006. El Niño-like pattern in the
Pacific during marine isotope stages (MIS) 13 and 11? Paleoceanography 21, PA1015.
Mollenhauer, G., Eglinton, T.I., Hopman, E.C., Sinninghe Damsté, J.S., 2008. A radiocarbonbased assessment of the preservation characteristics of crenarchaeol and alkenones from
continental margin sediments. Organic Geochemistry 39, 1039-1045.
Mollenhauer, G., Inthorn, M., Vogt, T., Zabel, M., Sinninghe Damsté, J.S., Eglinton, T.I.,
2007. Aging of marine organic matter during cross-shelf lateral transport in the Benguela
upwelling system revealed by compound-specific radiocarbon dating. Geochemistry
Geophysics Geosystems 8, Q09004.
Mollenhauer, G., McManus, J.F., Benthien, A., Müller, P.J., Eglinton, T.I., 2006. Rapid
lateral particle transport in the Argentine Basin: Molecular 14C and 230Thxs evidence. DeepSea Research Part I: Oceanographic Research Papers 53, 1224-1243.
Morales, C.E., Hormazabal, S.E., Blanco, J., 1999. Interannual variability in the mesoscale
distribution of the depth of the upper boundary of the oxygen minimum layer off northern
Chile (18–24°S): Implications for the pelagic system and biogeochemical cycling. Journal of
Marine Research 57, 909-932.
Müller, P.J., Cepek, M., Ruhland, G., Schneider, R.R., 1997. Alkenone and coccolithophorid
species changes in late Quaternary sediments from the Walvis Ridge: Implications for the
alkenone paleotemperature method. Palaeogeography, Palaeoclimatology, Palaeoecology 135,
71-96.
Müller, P.J., Kirst, G., Ruhland, G., von Storch, I., Rosell-Melé, A., 1998. Calibration of the
alkenone paleotemperature index U37K ' based on core-tops from the eastern South Atlantic
and the global ocean (60 °N-60 °S). Geochimica et Cosmochimica Acta 62, 1757-1772.
106 References Murray, A.E., Preston, C.M., Massana, R., Taylor, L.T., Blakis, A., Wu, K., DeLong, E.F.,
1998. Seasonal and spatial variability of bacterial and archaeal assemblages in the coastal
waters near Anvers Island, Antarctica. Applied and Environmental Microbiology 64, 25852595.
Naafs, B.D.A., Hefter, J., Stein, R., 2012. Application of the long chain diol index (LDI)
paleothermometer to the early Pleistocene (MIS 96). Organic Geochemistry 49, 83-85.
Nuernberg, D., 1995. Magnesium in tests of Neogloboquadrina pachyderma sinistral from
high northern and southern latitudes. The Journal of Foraminiferal Research 25, 350-368.
Orsi, A.H., Whitworth, T., Nowlin, W.D., 1995. On the meridional extent and fronts of the
Antarctic Circumpolar Current. Deep-Sea Research Part I: Oceanographic Research Papers
42, 641-673.
Pahnke, K., Zahn, R., Elderfield, H., Schulz, M., 2003. 340,000-Year Centennial-Scale
Marine Record of Southern Hemisphere Climatic Oscillation. Science 301, 948-952.
Paillard, D., Labeyrie, L., Yiou, P., 1996. Macintosh Program performs time-series analysis.
Eos Transactions AGU 77, 379-379.
Parrenin, F., Barnola, J.M., Beer, J., Blunier, T., Castellano, E., Chappellaz, J., Dreyfus, G.,
Fischer, H., Fujita, S., Jouzel, J., Kawamura, K., Lemieux-Dudon, B., Loulergue, L., MassonDelmotte, V., Narcisi, B., Petit, J.R., Raisbeck, G., Raynaud, D., Ruth, U., Schwander, J.,
Severi, M., Spahni, R., Steffensen, J.P., Svensson, A., Udisti, R., Waelbroeck, C., Wolff, E.,
2007. The EDC3 chronology for the EPICA dome C ice core. Climate of the Past 3, 485-497.
Peterse, F., Kim, J.H., Schouten, S., Kristensen, D.K., Koc, N., Sinninghe Damsté, J.S., 2009.
Constraints on the application of the MBT/CBT palaeothermometer at high latitude
environments (Svalbard, Norway). Organic Geochemistry 40, 692-699.
Petit, J.R., Jouzel, J., Raynaud, D., Barkov, N.I., Barnola, J.M., Basile, I., Bender, M.,
Chappellaz, J., Davis, M., Delaygue, G., Delmotte, M., Kotlyakov, V.M., Legrand, M.,
Lipenkov, V.Y., Lorius, C., Pepin, L., Ritz, C., Saltzman, E., Stievenard, M., 1999. Climate
and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica.
Nature 399, 429-436.
Pflaumann, U., Duprat, J., Pujol, C., Labeyrie, L.D., 1996. SIMMAX: A Modern Analog
Technique to Deduce Atlantic Sea Surface Temperatures from Planktonic Foraminifera in
Deep-Sea Sediments. Paleoceanography 11, 15-35.
Prahl, F.G., Mix, A.C., Sparrow, M.A., 2006. Alkenone paleothermometry: Biological lessons
from marine sediment records off western South America. Geochimica et Cosmochimica Acta
70, 101-117.
Prahl, F.G., Muehlhausen, L.A., Zahnle, D.L., 1988. Further evaluation of long-chain
alkenones as indicators of paleoceanographic conditions. Geochimica et Cosmochimica Acta
52, 2303-2310.
107 References Prahl, F.G., Wakeham, S.G., 1987. Calibration of unsaturation patterns in long-chain ketone
compositions for paleotemperature assessment. Nature 330, 367-369.
Prell, W.L., 1985. The stability of low-latitude sea-surface temperatures: An evaluation of the
CLIMAP reconstruction with emphasis on the positive SST anomalies,, US Department of
Energy, Washington, DC., pp. 1-60.
Quiñones, R.A., Levipan, H.A., Urrutia, H., 2009. Spatial and temporal variability of
planktonic archaeal abundance in the Humboldt Current System off Chile. Deep Sea Research
Part II: Topical Studies in Oceanography 56, 1073-1082.
Rampen, S.W., Willmott, V., Kim, J.-H., Uliana, E., Mollenhauer, G., Schefuß, E., Sinninghe
Damsté, J.S., Schouten, S., 2012. Long chain 1,13- and 1,15-diols as a potential proxy for
palaeotemperature reconstruction. Geochimica et Cosmochimica Acta 84, 204-216.
Rechka, J.A., Maxwell, J.R., 1988. Unusual long chain ketones of algal origin. Tetrahedron
Letters 29, 2599-2600.
Rincón-Martínez, D., Lamy, F., Contreras, S., Leduc, G., Bard, E., Saukel, C., Blanz, T.,
Mackensen, A., Tiedemann, R., 2010. More humid interglacials in Ecuador during the past
500 kyr linked to latitudinal shifts of the equatorial front and the Intertropical Convergence
Zone in the eastern tropical Pacific. Paleoceanography 25, PA2210.
Roemmich, D., John Gould, W., Gilson, J., 2012. 135 years of global ocean warming between
the Challenger expedition and the Argo Programme. Nature Climate Change 2, 425-428.
Romero, O.E., Kim, J.-H., Hebbeln, D., 2006. Paleoproductivity evolution off central Chile
from the Last Glacial Maximum to the Early Holocene. Quaternary Research 65, 519-525.
Rosell-Melé, A., 1998. Interhemispheric appraisal of the value of alkenone indices as
temperature and salinity proxies in high-latitude locations. Paleoceanography 13, 694-703.
Rosell-Melé, A., Carter, J., Eglinton, G., 1994. Distributions of long-chain alkenones and
alkyl alkenoates in marine surface sediments from the North East Atlantic. Organic
Geochemistry 22, 501-509.
Rosell-Melé, A., Comes, P., 1999. Evidence for a warm Last Glacial Maximum in the Nordic
seas or an example of shortcomings in UK'37 and UK37 to estimate low sea surface
temperature? Paleoceanography 14, 770-776.
Rosell-Melé, A., Eglinton, G., Pflaumann, U., Sarnthein, M., 1995. Atlantic core-top
calibration of the U37K index as a sea-surface palaeotemperature indicator. Geochimica et
Cosmochimica Acta 59, 3099-3107.
Rosell-Melé, A., Jansen, E., Weinelt, M., 2002. Appraisal of a molecular approach to infer
variations in surface ocean freshwater inputs into the North Atlantic during the last glacial.
Global and Planetary Change 34, 143-152.
108 References Rühlemann, C., Butzin, M., 2006. Alkenone temperature anomalies in the Brazil-Malvinas
Confluence area caused by lateral advection of suspended particulate material. Geochemistry
Geophysics Geosyststems 7, Q10015.
Saavedra-Pellitero, M., Flores, J.A., Lamy, F., Sierro, F.J., Cortina, A., 2011.
Coccolithophore estimates of paleotemperature and paleoproductivity changes in the
southeast Pacific over the past 27 kyr. Paleoceanography 26, PA1201.
Saukel, C., Lamy, F., Stuut, J.-B.W., Tiedemann, R., Vogt, C., 2011. Distribution and
provenance of wind-blown SE Pacific surface sediments. Marine Geology 280, 130-142.
Sawada, K., Shiraiwa, Y., 2004. Alkenone and alkenoic acid compositions of the membrane
fractions of Emiliania huxleyi. Phytochemistry 65, 1299-1307.
Schaefer, G., Rodger, J.S., Hayward, B.W., Kennett, J.P., Sabaa, A.T., Scott, G.H., 2005.
Planktic foraminiferal and sea surface temperature record during the last 1 Myr across the
Subtropical Front, Southwest Pacific. Marine Micropaleontology 54, 191-212.
Schlitzer, R., 2011. Ocean Data View, http://odv.awi.de.
Schneider-Mor, A., Yam, R., Bianchi, C., Kunz-Pirrung, M., Gersonde, R., Shemesh, A.,
2008. Nutrient regime at the siliceous belt of the Atlantic sector of the Southern Ocean during
the past 660 ka. Paleoceanography 23, PA3217.
Schneider, B., Leduc, G., Park, W., 2010. Disentangling seasonal signals in Holocene climate
trends by satellite-model-proxy integration. Paleoceanography 25, PA4217.
Schneider, R.R., Müller, P.J., Acheson, R., 1999. Atlantic alkenone sea-surface temperature
records: Low versus mid latitudes and differences between hemispheres, In: Abrantes, F.,
Mix, A.C. (Eds.), Reconstructing Ocean History: A Window into the Future. Kluwer
Academic / Plenum Publishers, New York, pp. 33-55.
Schouten, S., Hopmans, E.C., Pancost, R.D., Damsté, J.S.S., 2000. Widespread occurrence of
structurally diverse tetraether membrane lipids: Evidence for the ubiquitous presence of lowtemperature relatives of hyperthermophiles. Proceedings of the National Academy of Sciences
97, 14421-14426.
Schouten, S., Hopmans, E.C., Schefuss, E., Sinninghe Damsté, J.S., 2002. Distributional
variations in marine crenarchaeotal membrane lipids: a new tool for reconstructing ancient sea
water temperatures? Earth and Planetary Science Letters 204, 265-274.
Schouten, S., Hopmans, E.C., Sinninghe Damsté, J.S., 2004. The effect of maturity and
depositional redox conditions on archaeal tetraether lipid palaeothermometry. Organic
Geochemistry 35, 567-571.
Schouten, S., Huguet, C., Hopmans, E.C., Kienhuis, M.V., Sinninghe Damsté, J.S., 2007.
Analytical methodology for TEX86 paleothermometry by high-performance liquid
chromatography/atmospheric pressure chemical ionization-mass spectrometry. Analytical
Chemistry 79, 2940-2944.
109 References Schulz, H.-M., Schöner, A., Emeis, K.-C., 2000. Long-chain alkenone patterns in the Baltic
sea—an ocean-freshwater transition. Geochimica et Cosmochimica Acta 64, 469-477.
Shackleton, N.J., 1987. Oxygen isotopes, ice volume and sea level. Quaternary Science
Reviews 6, 183-190.
Shackleton, N.J., Berger, A., Peltier, W.R., 1990. An alternative astronomical calibration of
the lower Pleistocene timescale based on ODP Site 677. Earth and Environmental Science
Transactions of the Royal Society of Edinburgh 81, 251-261.
Shah, S.R., Mollenhauer, G., Ohkouchi, N., Eglinton, T.I., Pearson, A., 2008. Origins of
archaeal tetraether lipids in sediments: Insights from radiocarbon analysis. Geochimica et
Cosmochimica Acta 72, 4577-4594.
Shemesh, A., Charles, C.D., Fairbanks, R.G., 1992. Oxygen Isotopes in Biogenic Silica:
Global Changes in Ocean Temperature and Isotopic Composition. Science 256, 1434-1436.
Shevenell, A.E., Ingalls, A.E., Domack, E.W., Kelly, C., 2011. Holocene Southern Ocean
surface temperature variability west of the Antarctic Peninsula. Nature 470, 250-254.
Sikes, E.L., Sicre, M.A., 2002. Relationship of the tetra-unsaturated C37 alkenone to salinity
and temperature: Implications for paleoproxy applications. Geochemistry Geophysics
Geosystems 3, 1063.
Sikes, E.L., Volkman, J.K., 1993. Calibration of alkenone unsaturation ratios (Uk'37) for
paleotemperature estimation in cold polar waters. Geochimica et Cosmochimica Acta 57,
1883-1889.
Sikes, E.L., Volkman, J.K., Robertson, L.G., Pichon, J.J., 1997. Alkenones and alkenes in
surface waters and sediments of the Southern Ocean: Implications for paleotemperature
estimation in polar regions. Geochimica et Cosmochimica Acta 61, 1495-1505.
Sinninghe Damsté, J.S., Ossebaar, J., Abbas, B., Schouten, S., Verschuren, D., 2009. Fluxes
and distribution of tetraether lipids in an equatorial African lake: Constraints on the
application of the TEX86 palaeothermometer and BIT index in lacustrine settings. Geochimica
et Cosmochimica Acta 73, 4232-4249.
Sinninghe Damsté, J.S., Rijpstra, W.I.C., Hopmans, E.C., Prahl, F.G., Wakeham, S.G.,
Schouten, S., 2002a. Distribution of membrane lipids of planktonic Crenarchaeota in the
Arabian sea. Applied and Environmental Microbiology 68, 2997-3002.
Sinninghe Damsté, J.S., Schouten, S., Hopmans, E.C., van Duin, A.C.T., Geenevasen, J.A.J.,
2002b. Crenarchaeol: the characteristic core glycerol dibiphytanyl glycerol tetraether
membrane lipid of cosmopolitan pelagic crenarchaeota. Journal of Lipid Research 43, 16411651.
Sluijs, A., Schouten, S., Pagani, M., Woltering, M., Brinkhuis, H., Sinninghe Damsté, J.S.,
Dickens, G.R., Huber, M., Reichart, G.J., Stein, R., Matthiessen, J., Lourens, L.J.,
Pedentchouk, N., Backman, J., Moran, K., Scientists, E., 2006. Subtropical arctic ocean
temperatures during the Palaeocene/Eocene thermal maximum. Nature 441, 610-613.
110 References Smith, R.W., Bianchi, T.S., Li, X., 2012. A re-evaluation of the use of branched GDGTs as
terrestrial biomarkers: Implications for the BIT Index. Geochimica et Cosmochimica Acta 80,
14-29.
Sonzogni, C., Bard, E., Rostek, F., Lafont, R., Rosell-Melé, A., Eglinton, G., 1997. Core-top
calibration of the alkenone index vs sea surface temperature in the Indian Ocean. Deep-Sea
Research Part II- Topical Studies in Oceanography 44, 1445-1460.
Stephens, B.B., Keeling, R.F., 2000. The influence of Antarctic sea ice on glacial-interglacial
CO2 variations. Nature 404, 171-174.
Strub, P.T., Mesias, J.M., Montecino, V., Rutlant, J., Salinas, S., 1998. Coastal ocean
circulation off western South America. Wiley, New York.
Stuut, J.-B.W., Lamy, F., 2004. Climate variability at the southern boundaries of the Namib
(southwestern Africa) and Atacama (northern Chile) coastal deserts during the last 120,000
yr. Quaternary Research 62, 301-309.
Talley, L.D., Pickard, G.L., Emery, W.J., Swift, J.H., 2011. Descriptive Physical
Oceanograpgy: an Introduction, 6 ed. Academic Press.
Toggweiler, J.R., Russell, J., 2008. Ocean circulation in a warming climate. Nature 451, 286288.
Tomczak, M., Godfrey, J.S., 2003. Regional Oceanography: an Introduction, 2nd ed. Daya
Publishing House, Delhi.
Uda, I., Sugai, A., Itoh, Y.H., Itoh, T., 2001. Variation on molecular species of polar lipids
from Thermoplasma acidophilum depends on growth temperature. Lipids 36, 103-105.
Urey, H.C., 1947. The thermodynamic properties of isotopic substances. Journal of Chemical
Society, 562-581.
Venz, K.A., Hodell, D.A., 2002. New evidence for changes in Plio-Pleistocene deep water
circulation from Southern Ocean ODP Leg 177 Site 1090. Palaeogeography,
Palaeoclimatology, Palaeoecology 182, 197-220.
Verleye, T.J., Louwye, S., 2010. Late Quaternary environmental changes and latitudinal shifts
of the Antarctic Circumpolar Current as recorded by dinoflagellate cysts from offshore Chile
(41°S). Quaternary Science Reviews 29, 1025-1039.
Versteegh, G.J.M., Bosch, H.J., De Leeuw, J.W., 1997. Potential palaeoenvironmental
information of C24 to C36 mid-chain diols, keto-ols and mid-chain hydroxy fatty acids; a
critical review. Organic Geochemistry 27, 1-13.
Villanueva, J., Grimalt, J.O., 1996. Pitfalls in the chromatographic determination of the
alkenone Uk37 index for paleotemperature estimation. Journal of Chromatography A 723,
285-291.
111 References Waelbroeck, C., Labeyrie, L., Duplessy, J.C., Guiot, J., Labracherie, M., Leclaire, H., Duprat,
J., 1998. Improving Past Sea Surface Temperature Estimates Based on Planktonic Fossil
Faunas. Paleoceanography 13, 272-283.
Wakeham, S.G., Lewis, C.M., Hopmans, E.C., Schouten, S., Sinninghe Damsté, J.S., 2003.
Archaea mediate anaerobic oxidation of methane in deep euxinic waters of the Black Sea.
Geochimica et Cosmochimica Acta 67, 1359-1374.
Walsh, J.E., 1978. A data set on Northern Hemisphere sea ice extent, World Data Center-A
for Glaciology (Snow and Ice), "Glaciological Data, Report GD-2", pp. 49 - 51.
Weijers, J.W.H., Schouten, S., Spaargaren, O.C., Sinninghe Damsté, J.S., 2006. Occurrence
and distribution of tetraether membrane lipids in soils: Implications for the use of the TEX86
proxy and the BIT index. Organic Geochemistry 37, 1680-1693.
Wells, P., Okada, H., 1997. Response of nannoplankton to major changes in sea-surface
temperature and movements of hydrological fronts over Site DSDP 594 (south Chatham Rise,
southeastern New Zealand), during the last 130 kyr. Marine Micropaleontology 32, 341-363.
Wuchter, C., Schouten, S., Coolen, M.J.L., Sinninghe Damsté, J.S., 2004. Temperaturedependent variation in the distribution of tetraether membrane lipids of marine Crenarchaeota:
Implications for TEX86 paleothermometry. Paleoceanography 19, PA4028.
Wuchter, C., Schouten, S., Wakeham, S.G., Sinninghe Damsté, J.S., 2005. Temporal and
spatial variation in tetraether membrane lipids of marine Crenarchaeota in particulate organic
matter: Implications for TEX86 paleothermometry. Paleoceanography 20, PA3013.
Wuchter, C., Schouten, S., Wakeham, S.G., Sinninghe Damsté, J.S.S., 2006. Archaeal
tetraether membrane lipid fluxes in the northeastern Pacific and the Arabian Sea: Implications
for TEX86 paleothermometry. Paleoceanography 21, PA4208.
Wyrtki, K., 1965. Oceanography of the eastern equatorial Pacific Ocean. Oceanography and
Marine Biology 4, 33-68.
Zhu, C., Weijers, J.W.H., Wagner, T., Pan, J.M., Chen, J.F., Pancost, R.D., 2011. Sources and
distributions of tetraether lipids in surface sediments across a large river-dominated
continental margin. Organic Geochemistry 42, 376-386.
Zielinski, U., Gersonde, R., 2002. Plio–Pleistocene diatom biostratigraphy from ODP Leg
177, Atlantic sector of the Southern Ocean. Marine Micropaleontology 45, 225-268.
Zwally, H.J., Comiso, J.C., Parkinson, C.L., Campbell, W.J., Carsey, F.D., Gloersen, P.,
1983. Antarctic sea ice, 1973-1976: satellite passive-microwave observations, NASA Special
Publications SP-459. National Aeronatics and Space Administration, Washington, D.C., p.
206.
112 Data handling Data handling
All data presented in this thesis will be publicly available on the Pangaea database
(www.pangaea.de) once the manuscripts are published online.
Data from the manuscript in Chapter 3: http://doi.pangaea.de/10.1594/PANGAEA.792642
Data from the manuscript in Appendix 5: http://doi.pangaea.de/10.1594/PANGAEA.767415
113 Appendix 1 Appendix 1: Abbreviations and acronyms
ACC
AMJ
ANN
APCI
APF
ARGO
ASE
AVHRR
BIT
CHC
Cren’
EPICA
EEP
ENSO
EUC
FID
GAM
GC
GDGT
G-IG
HMS
HPLC
ICOADS
IKM
IPCC
IRD
ITCZ
IUPAC
JAS
JFM
LDI
LGM
LSR
MARGO
MAT
MBE
MIS
MS
MSD
NBS
NIOZ
NSIPP
ODV
OMZ
OND
PCA
PCC
Antarctic Circumpolar Current
April May June
Artificial Neural Network
Atmospheric pressure chemical ionization
Antarctic Polar Front
Array for Real-time Geostrophic Oceanography
Accelerated solvent extractor
Advanced Very High Resolution Radiometer
Branched and Isoprenoid Tetraether index
Cape Horn Current
Crenarchaeol regioisomer
European Project for Ice Coring in Antarctica
Eastern Equatorial Pacific
El Niño – Southern Oscillation
Equatorial Undercurrent
Flame ionization detector
Generalized additive model
Gas chromatography
Glycerol dialkyl glycerol tetraether
Glacial - interglacial
Her/His Majesty’s Ship
High performance liquid chromatography
International Comprehensive Atmosphere-Ocean Data Set
Imbrie and Kipp method
Intergovernmental Panel on Climate Change
Ice-rafted debris
Intertropical Convergence Zone
International Union of Pure and Applied Chemistry
July August September
January February March
Long-chain Diol Index
Last Glacial Maximum
Linear sedimentation rate
Multiproxy Approach for the Reconstruction of the Glacial Ocean surface
Modern analog technique
Mid Brunhes Event
Marine Isotope Stage
Mass spectrometer
Mass Selective Detector
National Bureau of Standards
Netherlands Institute for Sea Research
NASA (National Aeronautics and Space Adminitration) Seasonal-ToInterannual Prediction Project
Ocean Data View
Oxygen minimum zone
October November December
Principal component analysis
Peru-Chile Current
114 Appendix 1 PETM
PUC
RAM
SAF
SCAR
SEC
SIM
SIMMAX
SSI
SST
STF
TEX86
TEX86H
TEX86L
TF
UK37
UK’37
VPDB
WOA
WOCE
WOD
WSI
Paleocene-Eocene Thermal Maximum
Peru-Chile Undercurrent
Revised analog method
Subantarctic Front
Scientific Committee on Antarctic Science
South Equatorial Current
Selected ion monitoring
Modern analog technique with a similarity index
Summer sea ice extent
Sea surface temperature
Subtropical Front
TetraEther indeX with 86 carbon atoms
Modified TEX86; TetraEther indeX with 86 carbon atoms for reconstructing sea
surface temperature Higher 15°C
Modified TEX86; TetraEther indeX with 86 carbon atoms for reconstructing sea
surface temperature Lower than 15°C
Transfer function
Unsaturation of Ketone with 37 carbon atoms (isomers with two, three and
four double-bonds)
Simplified UK37; Unsaturation of Ketone with 37 carbon atoms (isomers with
two and three double-bonds)
Vienna Pee Dee Belemnite
World Ocean Atlas
World Ocean Circulation Experiment
World Ocean Data
Winter sea-ice extent
115 Appendix 2 Appendix 2: Units and chemical nomenclatures
%
‰
°N/S/W/E
°C
Ca
cm
DCM
KOH
kV
kyr
m
Mg
min
mL
mm
ms
mTorr
A
L
m
N2
ng
PTFE
r2
SiO
Sr
Sv
v/v
percent
per mil
degree North/South/West/East
degree Celsius
Calcium
centimeter (10-2 meter)
Dichloromethane
Potassium hydroxide
kilovolt (103 volt)
kiloyear (103 year)
meter
Magnesium
minute
milliliter (10-3 liter)
millimeter (10-3 meter)
millisecond (10-3 second)
millitorr
microampere (10-6 ampere)
microliter (10-6 liter)
micrometer (10-6 meter)
Nitrogen
nanogram (10-9 gram)
Polytetrafluoroethylene
coefficient of determination
Silica
Strontium
Sverdrup
volume / volume
116 Appendix 3 Appendix 3: List of figures
Figure 1.7
Major components of the climate system. (from Roger Pielke Sr.’s
www.pielkeclimatesci.wordpress.com).
A schematic of the framework for projecting future climate using
climate models (from Goosse et al.’s online textbook at
www.climate.be/textbook/).
Molecular structure and standard IUPAC names of C37 alkenones with
2, 3, and 4 double bonds (Figure from Marlowe et al. [1990],
following structure verification by Rechka and Maxwell [1988] and
double-bond position identification by de Leeuw et al. [1980]).
Molecular structure of isoprenoid glycerol dialkyl glycerol tetraether
(GDGT) lipids with different ring moieties (from Ho et al. [2011])
Literature review of paleo sea surface temperature reconstruction in
the subantarctic South Pacific south of 40°S.
Crossplots of GDGT-based indices, i.e., TEX86 and TEX86L, with
satellite SSTs from Kim et al. [2010].
Location of marine sediments used in this study.
Figure 1.8
Types of biomarkers and their precursors in the three domains of life.
17
Figure 1.9
Schematic of silica or alumina column chromatography to separate a
mixture of organic compounds into different fractions prior to
analysis. (Figure from explow.com/chromatography)
South Pacific core-top alkenone index values with and without
saponification
Correlation between the climatology SST data sets from the NSIPP
AVHRR Pathfinder and Erosion Global 9km [Casey and Cornillon,
1999] and the World Ocean Atlas 2009 [Locarnini et al., 2010].
Location of study sites in the global ocean.
18
Principal component analysis on the individual GDGTs in the
compilation of the data in this study and the published core-top data by
Kim et al. [2008], Kim et al. [2010] and Ho et al. [2011]
Relationship between the fractional abundance of individual GDGTs
and the WOA09 annual mean SSTs..
Correlation plots of TEX86 and TEX86L values with annual mean sea
surface temperature derived from World Ocean Atlas 2009 (WOA09).
Correlation plots of TEX86 and TEX 86L values with water temperatures
derived from World Ocean Atlas 2009 (WOA09) at various depths.
Correlation plots of TEX86 and TEX 86L values with mean seasonal sea
surface temperature derived from World Ocean Atlas 2009 (WOA09).
The residuals of temperature estimates derived from the latest global
core-top calibrations for TEX86 and TEX 86L [Kim et al., 2010].
Close-up of the residuals of TEX86 and TEX 86L derived SST estimates
relative to World Ocean Atlas 2009 (WOA09) summer SST in the
Arctic.
Compilation of the TEX86 and TEX86L data from this study and the
published data set of Kim et al. [2010], Kim et al. [2008], Leider et al.
30
Figure 1.1
Figure 1.2
Figure 1.3
Figure 1.4
Figure 1.5
Figure 1.6
Figure 1.10
Figure 1.11
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 2.5
Figure 2.6
Figure 2.7
Figure 2.8
Figure 2.9
1
2
5
8
11
12
16
19
21
27
32
33
35
36
37
40
44
117 Appendix 3 Figure 3.1
[2010], Ho et al. [2011] and Shevenell et al. [2011], as a function of
World Ocean Atlas 2009 (WOA09) annual mean SST.
Location of sites and major oceanic currents discussed in this work.
50
57
Figure 3.3
Age models and linear sedimentation rates at core sites GeoB 3388-1,
GeoB 3327-5 and PS75/034-2.
Correlations of alkenone indices, i.e., UK’37 and UK37 with temperature.
Figure 3.4
Planktic 18O and alkenone-based SST records.
60
Figure 3.5
Comparison of temperature records from the Southern Ocean and
Antarctica based on different proxies.
Meridional gradients of alkenone-inferred SSTs and mean annual
insolation along the Southeast Pacific and SST evolution in the
tropical Pacific.
Core locations of GeoB 3388-1 and TG7 [Calvo et al., 2001], and
major hydrographic features in the southeast Pacific.
Molecular structures of glycerol dialkyl glycerol tetraethers (GDGTs)
used for the calculation of the TEX86H and BIT indices.
Glycerol dialkyl glycerol tetraether (GDGT) parameters and benthic
18O record of core GeoB 3388-1.
Correlation between the Branched and Isoprenoid Tetraether (BIT)
index values and the TEX86H-derived SST estimates calculated with
the global core-top calibration of Kim et al. [2010].
Temperature records at site GeoB 3388-1.
66
Upper water column structure at site GeoB 3388-1 in terms of oxygen
concentration and temperature based on the annual mean
climatological data of WOA09
UK37 calibrations available in the literature.
83
Correlation between the sea surface salinity (SSS) and the sea surface
temperature (SST) at the core-top sites of PS75 reported in Chapter 3.
TEX86L-derived temperature residuals (difference between estimated
SST and climatological SST) by applying the current global
calibration of Kim et al. [2010] on published data sets [Ho et al., 2011;
Kim et al., 2010; Leider et al., 2010; Shevenell et al., 2011] and data
presented in Chapter 2.
Schematic of a possible shift in the global core-top calibrations of
TEX86 (currently calibrated against annual mean SST) after integrating
production and export information, which skew towards warm (cold)
seasons at high (low) latitudes.
92
Figure 3.2
Figure 3.6
Figure 4.1
Figure 4.2
Figure 4.3
Figure 4.4
Figure 4.5
Figure 4.6
Figure 5.1
Figure 5.2
Figure 5.3
Figure 5.4
58
68
75
76
77
79
80
91
94
94
118 Appendix 4 Appendix 4: List of tables
Table 1.1
Table 1.2
Table 1.3
Table 3.1
Overview of paleo sea surface temperature (SST) proxies.
Short description of several modified GDGT-based indices and
calibrations that deviate from the commonly applied global core-top
calibrations of Schouten et al. [2002], Kim et al. [2008] and Kim et al.
[2010].
Details of previous paleo sea surface temperature studies in the
subantarctic Southeast Pacific (south of 40°S).
Site information and alkenone index values of the Southern Ocean coretop samples retrieved via multicoring.
4
9
11
52
119 Appendix 5 Appendix 5: Additional first-authored manuscripts
“Alkenone paleothermometry based on the haptophyte algae”, in Encyclopedia of
Quaternary Science, Elsevier (in press)
By Ho, S.L., Naafs, B.D.A., and Lamy, F.

This review contains a comprehensive description of many aspects of the alkenone
paleothermometry, including a brief history of the development of this proxy,
chemical structures, genetic evolution of the source organism, the relationship
between the alkenone unsaturation and temperature in suspended matter, laboratory
cultures and sediments, and some secondary effects (such as non-thermal
physiological factors and degradation) that might affect the use of this
paleothermometry.
“Core top TEX86 values in the south and equatorial Pacific”, published in Organic
Geochemistry 42, pages 94-99 (2011)
By Ho, S.L., Yamamoto, M., Mollenhauer, G., and Minagawa, M.

This paper discusses whether there is a regional bias in the TEX86-SST relationship in
the Pacific, given that this ocean is under-represented in the global core-top calibration
of Kim et al. [2010]. The appraisal is based on the relative distribution of individual
GDGTs, the crossplots of the index values with atlas SSTs and the residuals of the
temperature estimates. The small data set presented here encompasses diverse oceanic
provinces (e.g., eastern equatorial Pacific cold tongue, west Pacific warm pool, the
Southern Ocean) from low to high latitudes.
120